includes/Parser.php

   1 <?php
   2 /**
   3  * File for Parser and related classes
   4  *
   5  * @addtogroup Parser
   6  */
   7
   8 /**
   9  * Update this version number when the ParserOutput format
  10  * changes in an incompatible way, so the parser cache
  11  * can automatically discard old data.
  12  */
  13 define( 'MW_PARSER_VERSION', '1.6.1' );
  14
  15 define( 'RLH_FOR_UPDATE', 1 );
  16
  17 # Allowed values for $mOutputType
  18 define( 'OT_HTML', 1 );
  19 define( 'OT_WIKI', 2 );
  20 define( 'OT_MSG' , 3 );
  21 define( 'OT_PREPROCESS', 4 );
  22
  23 # Flags for setFunctionHook
  24 define( 'SFH_NO_HASH', 1 );
  25
  26 # string parameter for extractTags which will cause it
  27 # to strip HTML comments in addition to regular
  28 # <XML>-style tags. This should not be anything we
  29 # may want to use in wikisyntax
  30 define( 'STRIP_COMMENTS', 'HTMLCommentStrip' );
  31
  32 # Constants needed for external link processing
  33 define( 'HTTP_PROTOCOLS', 'http:\/\/|https:\/\/' );
  34 # Everything except bracket, space, or control characters
  35 define( 'EXT_LINK_URL_CLASS', '[^][<>"\\x00-\\x20\\x7F]' );
  36 # Including space, but excluding newlines
  37 define( 'EXT_LINK_TEXT_CLASS', '[^\]\\x0a\\x0d]' );
  38 define( 'EXT_IMAGE_FNAME_CLASS', '[A-Za-z0-9_.,~%\\-+&;#*?!=()@\\x80-\\xFF]' );
  39 define( 'EXT_IMAGE_EXTENSIONS', 'gif|png|jpg|jpeg' );
  40 define( 'EXT_LINK_BRACKETED',  '/\[(\b(' . wfUrlProtocols() . ')'.
  41         EXT_LINK_URL_CLASS.'+) *('.EXT_LINK_TEXT_CLASS.'*?)\]/S' );
  42 define( 'EXT_IMAGE_REGEX',
  43         '/^('.HTTP_PROTOCOLS.')'.  # Protocol
  44         '('.EXT_LINK_URL_CLASS.'+)\\/'.  # Hostname and path
  45         '('.EXT_IMAGE_FNAME_CLASS.'+)\\.((?i)'.EXT_IMAGE_EXTENSIONS.')$/S' # Filename
  46 );
  47
  48 // State constants for the definition list colon extraction
  49 define( 'MW_COLON_STATE_TEXT', 0 );
  50 define( 'MW_COLON_STATE_TAG', 1 );
  51 define( 'MW_COLON_STATE_TAGSTART', 2 );
  52 define( 'MW_COLON_STATE_CLOSETAG', 3 );
  53 define( 'MW_COLON_STATE_TAGSLASH', 4 );
  54 define( 'MW_COLON_STATE_COMMENT', 5 );
  55 define( 'MW_COLON_STATE_COMMENTDASH', 6 );
  56 define( 'MW_COLON_STATE_COMMENTDASHDASH', 7 );
  57
  58 /**
  59  * PHP Parser
  60  *
  61  * Processes wiki markup
  62  *
  63  * <pre>
  64  * There are four main entry points into the Parser class:
  65  * parse()
  66  *   produces HTML output
  67  * preSaveTransform().
  68  *   produces altered wiki markup.
  69  * transformMsg()
  70  *   performs brace substitution on MediaWiki messages
  71  * preprocess()
  72  *   removes HTML comments and expands templates
  73  *
  74  * Globals used:
  75  *    objects:   $wgLang, $wgContLang
  76  *
  77  * NOT $wgArticle, $wgUser or $wgTitle. Keep them away!
  78  *
  79  * settings:
  80  *  $wgUseTex*, $wgUseDynamicDates*, $wgInterwikiMagic*,
  81  *  $wgNamespacesWithSubpages, $wgAllowExternalImages*,
  82  *  $wgLocaltimezone, $wgAllowSpecialInclusion*,
  83  *  $wgMaxArticleSize*
  84  *
  85  *  * only within ParserOptions
  86  * </pre>
  87  *
  88  */
  89 class Parser
  90 {
  91         const VERSION = MW_PARSER_VERSION;
  92         /**#@+
  93          * @private
  94          */
  95         # Persistent:
  96         var $mTagHooks, $mFunctionHooks, $mFunctionSynonyms, $mVariables;
  97
  98         # Cleared with clearState():
  99         var $mOutput, $mAutonumber, $mDTopen, $mStripState;
 100         var $mIncludeCount, $mArgStack, $mLastSection, $mInPre;
 101         var $mInterwikiLinkHolders, $mLinkHolders, $mUniqPrefix;
 102         var $mIncludeSizes, $mDefaultSort;
 103         var $mTemplates,        // cache of already loaded templates, avoids
 104                                 // multiple SQL queries for the same string
 105             $mTemplatePath;     // stores an unsorted hash of all the templates already loaded
 106                                 // in this path. Used for loop detection.
 107
 108         # Temporary
 109         # These are variables reset at least once per parse regardless of $clearState
 110         var $mOptions,      // ParserOptions object
 111                 $mTitle,        // Title context, used for self-link rendering and similar things
 112                 $mOutputType,   // Output type, one of the OT_xxx constants
 113                 $ot,            // Shortcut alias, see setOutputType()
 114                 $mRevisionId,   // ID to display in {{REVISIONID}} tags
 115                 $mRevisionTimestamp, // The timestamp of the specified revision ID
 116                 $mRevIdForTs;   // The revision ID which was used to fetch the timestamp
 117
 118         /**#@-*/
 119
 120         /**
 121          * Constructor
 122          *
 123          * @public
 124          */
 125         function Parser() {
 126                 $this->mTagHooks = array();
 127                 $this->mFunctionHooks = array();
 128                 $this->mFunctionSynonyms = array( 0 => array(), 1 => array() );
 129                 $this->mFirstCall = true;
 130         }
 131
 132         /**
 133          * Do various kinds of initialisation on the first call of the parser
 134          */
 135         function firstCallInit() {
 136                 if ( !$this->mFirstCall ) {
 137                         return;
 138                 }
 139
 140                 wfProfileIn( __METHOD__ );
 141                 global $wgAllowDisplayTitle, $wgAllowSlowParserFunctions;
 142
 143                 $this->setHook( 'pre', array( $this, 'renderPreTag' ) );
 144
 145                 $this->setFunctionHook( 'int', array( 'CoreParserFunctions', 'intFunction' ), SFH_NO_HASH );
 146                 $this->setFunctionHook( 'ns', array( 'CoreParserFunctions', 'ns' ), SFH_NO_HASH );
 147                 $this->setFunctionHook( 'urlencode', array( 'CoreParserFunctions', 'urlencode' ), SFH_NO_HASH );
 148                 $this->setFunctionHook( 'lcfirst', array( 'CoreParserFunctions', 'lcfirst' ), SFH_NO_HASH );
 149                 $this->setFunctionHook( 'ucfirst', array( 'CoreParserFunctions', 'ucfirst' ), SFH_NO_HASH );
 150                 $this->setFunctionHook( 'lc', array( 'CoreParserFunctions', 'lc' ), SFH_NO_HASH );
 151                 $this->setFunctionHook( 'uc', array( 'CoreParserFunctions', 'uc' ), SFH_NO_HASH );
 152                 $this->setFunctionHook( 'localurl', array( 'CoreParserFunctions', 'localurl' ), SFH_NO_HASH );
 153                 $this->setFunctionHook( 'localurle', array( 'CoreParserFunctions', 'localurle' ), SFH_NO_HASH );
 154                 $this->setFunctionHook( 'fullurl', array( 'CoreParserFunctions', 'fullurl' ), SFH_NO_HASH );
 155                 $this->setFunctionHook( 'fullurle', array( 'CoreParserFunctions', 'fullurle' ), SFH_NO_HASH );
 156                 $this->setFunctionHook( 'formatnum', array( 'CoreParserFunctions', 'formatnum' ), SFH_NO_HASH );
 157                 $this->setFunctionHook( 'grammar', array( 'CoreParserFunctions', 'grammar' ), SFH_NO_HASH );
 158                 $this->setFunctionHook( 'plural', array( 'CoreParserFunctions', 'plural' ), SFH_NO_HASH );
 159                 $this->setFunctionHook( 'numberofpages', array( 'CoreParserFunctions', 'numberofpages' ), SFH_NO_HASH );
 160                 $this->setFunctionHook( 'numberofusers', array( 'CoreParserFunctions', 'numberofusers' ), SFH_NO_HASH );
 161                 $this->setFunctionHook( 'numberofarticles', array( 'CoreParserFunctions', 'numberofarticles' ), SFH_NO_HASH );
 162                 $this->setFunctionHook( 'numberoffiles', array( 'CoreParserFunctions', 'numberoffiles' ), SFH_NO_HASH );
 163                 $this->setFunctionHook( 'numberofadmins', array( 'CoreParserFunctions', 'numberofadmins' ), SFH_NO_HASH );
 164                 $this->setFunctionHook( 'language', array( 'CoreParserFunctions', 'language' ), SFH_NO_HASH );
 165                 $this->setFunctionHook( 'padleft', array( 'CoreParserFunctions', 'padleft' ), SFH_NO_HASH );
 166                 $this->setFunctionHook( 'padright', array( 'CoreParserFunctions', 'padright' ), SFH_NO_HASH );
 167                 $this->setFunctionHook( 'anchorencode', array( 'CoreParserFunctions', 'anchorencode' ), SFH_NO_HASH );
 168                 $this->setFunctionHook( 'special', array( 'CoreParserFunctions', 'special' ) );
 169                 $this->setFunctionHook( 'defaultsort', array( 'CoreParserFunctions', 'defaultsort' ), SFH_NO_HASH );
 170
 171                 if ( $wgAllowDisplayTitle ) {
 172                         $this->setFunctionHook( 'displaytitle', array( 'CoreParserFunctions', 'displaytitle' ), SFH_NO_HASH );
 173                 }
 174                 if ( $wgAllowSlowParserFunctions ) {
 175                         $this->setFunctionHook( 'pagesinnamespace', array( 'CoreParserFunctions', 'pagesinnamespace' ), SFH_NO_HASH );
 176                 }
 177
 178                 $this->initialiseVariables();
 179                 $this->mFirstCall = false;
 180                 wfProfileOut( __METHOD__ );
 181         }
 182
 183         /**
 184          * Clear Parser state
 185          *
 186          * @private
 187          */
 188         function clearState() {
 189                 wfProfileIn( __METHOD__ );
 190                 if ( $this->mFirstCall ) {
 191                         $this->firstCallInit();
 192                 }
 193                 $this->mOutput = new ParserOutput;
 194                 $this->mAutonumber = 0;
 195                 $this->mLastSection = '';
 196                 $this->mDTopen = false;
 197                 $this->mIncludeCount = array();
 198                 $this->mStripState = new StripState;
 199                 $this->mArgStack = array();
 200                 $this->mInPre = false;
 201                 $this->mInterwikiLinkHolders = array(
 202                         'texts' => array(),
 203                         'titles' => array()
 204                 );
 205                 $this->mLinkHolders = array(
 206                         'namespaces' => array(),
 207                         'dbkeys' => array(),
 208                         'queries' => array(),
 209                         'texts' => array(),
 210                         'titles' => array()
 211                 );
 212                 $this->mRevisionTimestamp = $this->mRevisionId = null;
 213
 214                 /**
 215                  * Prefix for temporary replacement strings for the multipass parser.
 216                  * \x07 should never appear in input as it's disallowed in XML.
 217                  * Using it at the front also gives us a little extra robustness
 218                  * since it shouldn't match when butted up against identifier-like
 219                  * string constructs.
 220                  */
 221                 $this->mUniqPrefix = "\x07UNIQ" . Parser::getRandomString();
 222
 223                 # Clear these on every parse, bug 4549
 224                 $this->mTemplates = array();
 225                 $this->mTemplatePath = array();
 226
 227                 $this->mShowToc = true;
 228                 $this->mForceTocPosition = false;
 229                 $this->mIncludeSizes = array(
 230                         'pre-expand' => 0,
 231                         'post-expand' => 0,
 232                         'arg' => 0
 233                 );
 234                 $this->mDefaultSort = false;
 235
 236                 wfRunHooks( 'ParserClearState', array( &$this ) );
 237                 wfProfileOut( __METHOD__ );
 238         }
 239
 240         function setOutputType( $ot ) {
 241                 $this->mOutputType = $ot;
 242                 // Shortcut alias
 243                 $this->ot = array(
 244                         'html' => $ot == OT_HTML,
 245                         'wiki' => $ot == OT_WIKI,
 246                         'msg' => $ot == OT_MSG,
 247                         'pre' => $ot == OT_PREPROCESS,
 248                 );
 249         }
 250
 251         /**
 252          * Accessor for mUniqPrefix.
 253          *
 254          * @public
 255          */
 256         function uniqPrefix() {
 257                 return $this->mUniqPrefix;
 258         }
 259
 260         /**
 261          * Convert wikitext to HTML
 262          * Do not call this function recursively.
 263          *
 264          * @param string $text Text we want to parse
 265          * @param Title &$title A title object
 266          * @param array $options
 267          * @param boolean $linestart
 268          * @param boolean $clearState
 269          * @param int $revid number to pass in {{REVISIONID}}
 270          * @return ParserOutput a ParserOutput
 271          */
 272         public function parse( $text, &$title, $options, $linestart = true, $clearState = true, $revid = null ) {
 273                 /**
 274                  * First pass--just handle <nowiki> sections, pass the rest off
 275                  * to internalParse() which does all the real work.
 276                  */
 277
 278                 global $wgUseTidy, $wgAlwaysUseTidy, $wgContLang;
 279                 $fname = 'Parser::parse-' . wfGetCaller();
 280                 wfProfileIn( __METHOD__ );
 281                 wfProfileIn( $fname );
 282
 283                 if ( $clearState ) {
 284                         $this->clearState();
 285                 }
 286
 287                 $this->mOptions = $options;
 288                 $this->mTitle =& $title;
 289                 $oldRevisionId = $this->mRevisionId;
 290                 $oldRevisionTimestamp = $this->mRevisionTimestamp;
 291                 if( $revid !== null ) {
 292                         $this->mRevisionId = $revid;
 293                         $this->mRevisionTimestamp = null;
 294                 }
 295                 $this->setOutputType( OT_HTML );
 296                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 297                 $text = $this->strip( $text, $this->mStripState );
 298                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 299                 $text = $this->internalParse( $text );
 300                 $text = $this->mStripState->unstripGeneral( $text );
 301
 302                 # Clean up special characters, only run once, next-to-last before doBlockLevels
 303                 $fixtags = array(
 304                         # french spaces, last one Guillemet-left
 305                         # only if there is something before the space
 306                         '/(.) (?=\\?|:|;|!|\\302\\273)/' => '\\1&nbsp;\\2',
 307                         # french spaces, Guillemet-right
 308                         '/(\\302\\253) /' => '\\1&nbsp;',
 309                 );
 310                 $text = preg_replace( array_keys($fixtags), array_values($fixtags), $text );
 311
 312                 # only once and last
 313                 $text = $this->doBlockLevels( $text, $linestart );
 314
 315                 $this->replaceLinkHolders( $text );
 316
 317                 # the position of the parserConvert() call should not be changed. it
 318                 # assumes that the links are all replaced and the only thing left
 319                 # is the <nowiki> mark.
 320                 # Side-effects: this calls $this->mOutput->setTitleText()
 321                 $text = $wgContLang->parserConvert( $text, $this );
 322
 323                 $text = $this->mStripState->unstripNoWiki( $text );
 324
 325                 wfRunHooks( 'ParserBeforeTidy', array( &$this, &$text ) );
 326
 327                 $text = Sanitizer::normalizeCharReferences( $text );
 328
 329                 if (($wgUseTidy and $this->mOptions->mTidy) or $wgAlwaysUseTidy) {
 330                         $text = Parser::tidy($text);
 331                 } else {
 332                         # attempt to sanitize at least some nesting problems
 333                         # (bug #2702 and quite a few others)
 334                         $tidyregs = array(
 335                                 # ''Something [http://www.cool.com cool''] -->
 336                                 # <i>Something</i><a href="http://www.cool.com"..><i>cool></i></a>
 337                                 '/(<([bi])>)(<([bi])>)?([^<]*)(<\/?a[^<]*>)([^<]*)(<\/\\4>)?(<\/\\2>)/' =>
 338                                 '\\1\\3\\5\\8\\9\\6\\1\\3\\7\\8\\9',
 339                                 # fix up an anchor inside another anchor, only
 340                                 # at least for a single single nested link (bug 3695)
 341                                 '/(<a[^>]+>)([^<]*)(<a[^>]+>[^<]*)<\/a>(.*)<\/a>/' =>
 342                                 '\\1\\2</a>\\3</a>\\1\\4</a>',
 343                                 # fix div inside inline elements- doBlockLevels won't wrap a line which
 344                                 # contains a div, so fix it up here; replace
 345                                 # div with escaped text
 346                                 '/(<([aib]) [^>]+>)([^<]*)(<div([^>]*)>)(.*)(<\/div>)([^<]*)(<\/\\2>)/' =>
 347                                 '\\1\\3&lt;div\\5&gt;\\6&lt;/div&gt;\\8\\9',
 348                                 # remove empty italic or bold tag pairs, some
 349                                 # introduced by rules above
 350                                 '/<([bi])><\/\\1>/' => '',
 351                         );
 352
 353                         $text = preg_replace(
 354                                 array_keys( $tidyregs ),
 355                                 array_values( $tidyregs ),
 356                                 $text );
 357                 }
 358
 359                 wfRunHooks( 'ParserAfterTidy', array( &$this, &$text ) );
 360
 361                 # Information on include size limits, for the benefit of users who try to skirt them
 362                 if ( max( $this->mIncludeSizes ) > 1000 ) {
 363                         $max = $this->mOptions->getMaxIncludeSize();
 364                         $text .= "<!-- \n" .
 365                                 "Pre-expand include size: {$this->mIncludeSizes['pre-expand']} bytes\n" .
 366                                 "Post-expand include size: {$this->mIncludeSizes['post-expand']} bytes\n" .
 367                                 "Template argument size: {$this->mIncludeSizes['arg']} bytes\n" .
 368                                 "Maximum: $max bytes\n" .
 369                                 "-->\n";
 370                 }
 371                 $this->mOutput->setText( $text );
 372                 $this->mRevisionId = $oldRevisionId;
 373                 $this->mRevisionTimestamp = $oldRevisionTimestamp;
 374                 wfProfileOut( $fname );
 375                 wfProfileOut( __METHOD__ );
 376
 377                 return $this->mOutput;
 378         }
 379
 380         /**
 381          * Recursive parser entry point that can be called from an extension tag
 382          * hook.
 383          */
 384         function recursiveTagParse( $text ) {
 385                 wfProfileIn( __METHOD__ );
 386                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 387                 $text = $this->strip( $text, $this->mStripState );
 388                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 389                 $text = $this->internalParse( $text );
 390                 wfProfileOut( __METHOD__ );
 391                 return $text;
 392         }
 393
 394         /**
 395          * Expand templates and variables in the text, producing valid, static wikitext.
 396          * Also removes comments.
 397          */
 398         function preprocess( $text, $title, $options ) {
 399                 wfProfileIn( __METHOD__ );
 400                 $this->clearState();
 401                 $this->setOutputType( OT_PREPROCESS );
 402                 $this->mOptions = $options;
 403                 $this->mTitle = $title;
 404                 wfRunHooks( 'ParserBeforeStrip', array( &$this, &$text, &$this->mStripState ) );
 405                 $text = $this->strip( $text, $this->mStripState );
 406                 wfRunHooks( 'ParserAfterStrip', array( &$this, &$text, &$this->mStripState ) );
 407                 if ( $this->mOptions->getRemoveComments() ) {
 408                         $text = Sanitizer::removeHTMLcomments( $text );
 409                 }
 410                 $text = $this->replaceVariables( $text );
 411                 $text = $this->mStripState->unstripBoth( $text );
 412                 wfProfileOut( __METHOD__ );
 413                 return $text;
 414         }
 415
 416         /**
 417          * Get a random string
 418          *
 419          * @private
 420          * @static
 421          */
 422         function getRandomString() {
 423                 return dechex(mt_rand(0, 0x7fffffff)) . dechex(mt_rand(0, 0x7fffffff));
 424         }
 425
 426         function &getTitle() { return $this->mTitle; }
 427         function getOptions() { return $this->mOptions; }
 428
 429         function getFunctionLang() {
 430                 global $wgLang, $wgContLang;
 431                 return $this->mOptions->getInterfaceMessage() ? $wgLang : $wgContLang;
 432         }
 433
 434         /**
 435          * Replaces all occurrences of HTML-style comments and the given tags
 436          * in the text with a random marker and returns teh next text. The output
 437          * parameter $matches will be an associative array filled with data in
 438          * the form:
 439          *   'UNIQ-xxxxx' => array(
 440          *     'element',
 441          *     'tag content',
 442          *     array( 'param' => 'x' ),
 443          *     '<element param="x">tag content</element>' ) )
 444          *
 445          * @param $elements list of element names. Comments are always extracted.
 446          * @param $text Source text string.
 447          * @param $uniq_prefix
 448          *
 449          * @private
 450          * @static
 451          */
 452         function extractTagsAndParams($elements, $text, &$matches, $uniq_prefix = ''){
 453                 static $n = 1;
 454                 $stripped = '';
 455                 $matches = array();
 456
 457                 $taglist = implode( '|', $elements );
 458                 $start = "/<($taglist)(\\s+[^>]*?|\\s*?)(\/?>)|<(!--)/i";
 459
 460                 while ( '' != $text ) {
 461                         $p = preg_split( $start, $text, 2, PREG_SPLIT_DELIM_CAPTURE );
 462                         $stripped .= $p[0];
 463                         if( count( $p ) < 5 ) {
 464                                 break;
 465                         }
 466                         if( count( $p ) > 5 ) {
 467                                 // comment
 468                                 $element    = $p[4];
 469                                 $attributes = '';
 470                                 $close      = '';
 471                                 $inside     = $p[5];
 472                         } else {
 473                                 // tag
 474                                 $element    = $p[1];
 475                                 $attributes = $p[2];
 476                                 $close      = $p[3];
 477                                 $inside     = $p[4];
 478                         }
 479
 480                         $marker = "$uniq_prefix-$element-" . sprintf('%08X', $n++) . '-QINU';
 481                         $stripped .= $marker;
 482
 483                         if ( $close === '/>' ) {
 484                                 // Empty element tag, <tag />
 485                                 $content = null;
 486                                 $text = $inside;
 487                                 $tail = null;
 488                         } else {
 489                                 if( $element == '!--' ) {
 490                                         $end = '/(-->)/';
 491                                 } else {
 492                                         $end = "/(<\\/$element\\s*>)/i";
 493                                 }
 494                                 $q = preg_split( $end, $inside, 2, PREG_SPLIT_DELIM_CAPTURE );
 495                                 $content = $q[0];
 496                                 if( count( $q ) < 3 ) {
 497                                         # No end tag -- let it run out to the end of the text.
 498                                         $tail = '';
 499                                         $text = '';
 500                                 } else {
 501                                         $tail = $q[1];
 502                                         $text = $q[2];
 503                                 }
 504                         }
 505
 506                         $matches[$marker] = array( $element,
 507                                 $content,
 508                                 Sanitizer::decodeTagAttributes( $attributes ),
 509                                 "<$element$attributes$close$content$tail" );
 510                 }
 511                 return $stripped;
 512         }
 513
 514         /**
 515          * Strips and renders nowiki, pre, math, hiero
 516          * If $render is set, performs necessary rendering operations on plugins
 517          * Returns the text, and fills an array with data needed in unstrip()
 518          *
 519          * @param StripState $state
 520          *
 521          * @param bool $stripcomments when set, HTML comments <!-- like this -->
 522          *  will be stripped in addition to other tags. This is important
 523          *  for section editing, where these comments cause confusion when
 524          *  counting the sections in the wikisource
 525          *
 526          * @param array dontstrip contains tags which should not be stripped;
 527          *  used to prevent stipping of <gallery> when saving (fixes bug 2700)
 528          *
 529          * @private
 530          */
 531         function strip( $text, $state, $stripcomments = false , $dontstrip = array () ) {
 532                 global $wgContLang;
 533                 wfProfileIn( __METHOD__ );
 534                 $render = ($this->mOutputType == OT_HTML);
 535
 536                 $uniq_prefix = $this->mUniqPrefix;
 537                 $commentState = new ReplacementArray;
 538                 $nowikiItems = array();
 539                 $generalItems = array();
 540
 541                 $elements = array_merge(
 542                         array( 'nowiki', 'gallery' ),
 543                         array_keys( $this->mTagHooks ) );
 544                 global $wgRawHtml;
 545                 if( $wgRawHtml ) {
 546                         $elements[] = 'html';
 547                 }
 548                 if( $this->mOptions->getUseTeX() ) {
 549                         $elements[] = 'math';
 550                 }
 551
 552                 # Removing $dontstrip tags from $elements list (currently only 'gallery', fixing bug 2700)
 553                 foreach ( $elements AS $k => $v ) {
 554                         if ( !in_array ( $v , $dontstrip ) ) continue;
 555                         unset ( $elements[$k] );
 556                 }
 557
 558                 $matches = array();
 559                 $text = Parser::extractTagsAndParams( $elements, $text, $matches, $uniq_prefix );
 560
 561                 foreach( $matches as $marker => $data ) {
 562                         list( $element, $content, $params, $tag ) = $data;
 563                         if( $render ) {
 564                                 $tagName = strtolower( $element );
 565                                 wfProfileIn( __METHOD__."-render-$tagName" );
 566                                 switch( $tagName ) {
 567                                 case '!--':
 568                                         // Comment
 569                                         if( substr( $tag, -3 ) == '-->' ) {
 570                                                 $output = $tag;
 571                                         } else {
 572                                                 // Unclosed comment in input.
 573                                                 // Close it so later stripping can remove it
 574                                                 $output = "$tag-->";
 575                                         }
 576                                         break;
 577                                 case 'html':
 578                                         if( $wgRawHtml ) {
 579                                                 $output = $content;
 580                                                 break;
 581                                         }
 582                                         // Shouldn't happen otherwise. :)
 583                                 case 'nowiki':
 584                                         $output = Xml::escapeTagsOnly( $content );
 585                                         break;
 586                                 case 'math':
 587                                         $output = $wgContLang->armourMath( MathRenderer::renderMath( $content ) );
 588                                         break;
 589                                 case 'gallery':
 590                                         $output = $this->renderImageGallery( $content, $params );
 591                                         break;
 592                                 default:
 593                                         if( isset( $this->mTagHooks[$tagName] ) ) {
 594                                                 $output = call_user_func_array( $this->mTagHooks[$tagName],
 595                                                         array( $content, $params, $this ) );
 596                                         } else {
 597                                                 throw new MWException( "Invalid call hook $element" );
 598                                         }
 599                                 }
 600                                 wfProfileOut( __METHOD__."-render-$tagName" );
 601                         } else {
 602                                 // Just stripping tags; keep the source
 603                                 $output = $tag;
 604                         }
 605
 606                         // Unstrip the output, to support recursive strip() calls
 607                         $output = $state->unstripBoth( $output );
 608
 609                         if( !$stripcomments && $element == '!--' ) {
 610                                 $commentState->setPair( $marker, $output );
 611                         } elseif ( $element == 'html' || $element == 'nowiki' ) {
 612                                 $nowikiItems[$marker] = $output;
 613                         } else {
 614                                 $generalItems[$marker] = $output;
 615                         }
 616                 }
 617                 # Add the new items to the state
 618                 # We do this after the loop instead of during it to avoid slowing
 619                 # down the recursive unstrip
 620                 $state->nowiki->mergeArray( $nowikiItems );
 621                 $state->general->mergeArray( $generalItems );
 622
 623                 # Unstrip comments unless explicitly told otherwise.
 624                 # (The comments are always stripped prior to this point, so as to
 625                 # not invoke any extension tags / parser hooks contained within
 626                 # a comment.)
 627                 if ( !$stripcomments ) {
 628                         // Put them all back and forget them
 629                         $text = $commentState->replace( $text );
 630                 }
 631
 632                 wfProfileOut( __METHOD__ );
 633                 return $text;
 634         }
 635
 636         /**
 637          * Restores pre, math, and other extensions removed by strip()
 638          *
 639          * always call unstripNoWiki() after this one
 640          * @private
 641          * @deprecated use $this->mStripState->unstrip()
 642          */
 643         function unstrip( $text, $state ) {
 644                 return $state->unstripGeneral( $text );
 645         }
 646
 647         /**
 648          * Always call this after unstrip() to preserve the order
 649          *
 650          * @private
 651          * @deprecated use $this->mStripState->unstrip()
 652          */
 653         function unstripNoWiki( $text, $state ) {
 654                 return $state->unstripNoWiki( $text );
 655         }
 656
 657         /**
 658          * @deprecated use $this->mStripState->unstripBoth()
 659          */
 660         function unstripForHTML( $text ) {
 661                 return $this->mStripState->unstripBoth( $text );
 662         }
 663
 664         /**
 665          * Add an item to the strip state
 666          * Returns the unique tag which must be inserted into the stripped text
 667          * The tag will be replaced with the original text in unstrip()
 668          *
 669          * @private
 670          */
 671         function insertStripItem( $text, &$state ) {
 672                 $rnd = $this->mUniqPrefix . '-item' . Parser::getRandomString();
 673                 $state->general->setPair( $rnd, $text );
 674                 return $rnd;
 675         }
 676
 677         /**
 678          * Interface with html tidy, used if $wgUseTidy = true.
 679          * If tidy isn't able to correct the markup, the original will be
 680          * returned in all its glory with a warning comment appended.
 681          *
 682          * Either the external tidy program or the in-process tidy extension
 683          * will be used depending on availability. Override the default
 684          * $wgTidyInternal setting to disable the internal if it's not working.
 685          *
 686          * @param string $text Hideous HTML input
 687          * @return string Corrected HTML output
 688          * @public
 689          * @static
 690          */
 691         function tidy( $text ) {
 692                 global $wgTidyInternal;
 693                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
 694 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
 695 '<head><title>test</title></head><body>'.$text.'</body></html>';
 696                 if( $wgTidyInternal ) {
 697                         $correctedtext = Parser::internalTidy( $wrappedtext );
 698                 } else {
 699                         $correctedtext = Parser::externalTidy( $wrappedtext );
 700                 }
 701                 if( is_null( $correctedtext ) ) {
 702                         wfDebug( "Tidy error detected!\n" );
 703                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 704                 }
 705                 return $correctedtext;
 706         }
 707
 708         /**
 709          * Spawn an external HTML tidy process and get corrected markup back from it.
 710          *
 711          * @private
 712          * @static
 713          */
 714         function externalTidy( $text ) {
 715                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 716                 $fname = 'Parser::externalTidy';
 717                 wfProfileIn( $fname );
 718
 719                 $cleansource = '';
 720                 $opts = ' -utf8';
 721
 722                 $descriptorspec = array(
 723                         0 => array('pipe', 'r'),
 724                         1 => array('pipe', 'w'),
 725                         2 => array('file', '/dev/null', 'a')  // FIXME: this line in UNIX-specific, it generates a warning on Windows, because /dev/null is not a valid Windows file.
 726                 );
 727                 $pipes = array();
 728                 $process = proc_open("$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes);
 729                 if (is_resource($process)) {
 730                         // Theoretically, this style of communication could cause a deadlock
 731                         // here. If the stdout buffer fills up, then writes to stdin could
 732                         // block. This doesn't appear to happen with tidy, because tidy only
 733                         // writes to stdout after it's finished reading from stdin. Search
 734                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 735                         fwrite($pipes[0], $text);
 736                         fclose($pipes[0]);
 737                         while (!feof($pipes[1])) {
 738                                 $cleansource .= fgets($pipes[1], 1024);
 739                         }
 740                         fclose($pipes[1]);
 741                         proc_close($process);
 742                 }
 743
 744                 wfProfileOut( $fname );
 745
 746                 if( $cleansource == '' && $text != '') {
 747                         // Some kind of error happened, so we couldn't get the corrected text.
 748                         // Just give up; we'll use the source text and append a warning.
 749                         return null;
 750                 } else {
 751                         return $cleansource;
 752                 }
 753         }
 754
 755         /**
 756          * Use the HTML tidy PECL extension to use the tidy library in-process,
 757          * saving the overhead of spawning a new process. Currently written to
 758          * the PHP 4.3.x version of the extension, may not work on PHP 5.
 759          *
 760          * 'pear install tidy' should be able to compile the extension module.
 761          *
 762          * @private
 763          * @static
 764          */
 765         function internalTidy( $text ) {
 766                 global $wgTidyConf;
 767                 $fname = 'Parser::internalTidy';
 768                 wfProfileIn( $fname );
 769
 770                 tidy_load_config( $wgTidyConf );
 771                 tidy_set_encoding( 'utf8' );
 772                 tidy_parse_string( $text );
 773                 tidy_clean_repair();
 774                 if( tidy_get_status() == 2 ) {
 775                         // 2 is magic number for fatal error
 776                         // http://www.php.net/manual/en/function.tidy-get-status.php
 777                         $cleansource = null;
 778                 } else {
 779                         $cleansource = tidy_get_output();
 780                 }
 781                 wfProfileOut( $fname );
 782                 return $cleansource;
 783         }
 784
 785         /**
 786          * parse the wiki syntax used to render tables
 787          *
 788          * @private
 789          */
 790         function doTableStuff ( $text ) {
 791                 $fname = 'Parser::doTableStuff';
 792                 wfProfileIn( $fname );
 793
 794                 $lines = explode ( "\n" , $text );
 795                 $td_history = array (); // Is currently a td tag open?
 796                 $last_tag_history = array (); // Save history of last lag activated (td, th or caption)
 797                 $tr_history = array (); // Is currently a tr tag open?
 798                 $tr_attributes = array (); // history of tr attributes
 799                 $has_opened_tr = array(); // Did this table open a <tr> element?
 800                 $indent_level = 0; // indent level of the table
 801                 foreach ( $lines as $key => $line )
 802                 {
 803                         $line = trim ( $line );
 804
 805                         if( $line == '' ) { // empty line, go to next line
 806                                 continue;
 807                         }
 808                         $first_character = $line{0};
 809                         $matches = array();
 810
 811                         if ( preg_match( '/^(:*)\{\|(.*)$/' , $line , $matches ) ) {
 812                                 // First check if we are starting a new table
 813                                 $indent_level = strlen( $matches[1] );
 814
 815                                 $attributes = $this->mStripState->unstripBoth( $matches[2] );
 816                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'table' );
 817
 818                                 $lines[$key] = str_repeat( '<dl><dd>' , $indent_level ) . "<table{$attributes}>";
 819                                 array_push ( $td_history , false );
 820                                 array_push ( $last_tag_history , '' );
 821                                 array_push ( $tr_history , false );
 822                                 array_push ( $tr_attributes , '' );
 823                                 array_push ( $has_opened_tr , false );
 824                         } else if ( count ( $td_history ) == 0 ) {
 825                                 // Don't do any of the following
 826                                 continue;
 827                         } else if ( substr ( $line , 0 , 2 ) == '|}' ) {
 828                                 // We are ending a table
 829                                 $line = '</table>' . substr ( $line , 2 );
 830                                 $last_tag = array_pop ( $last_tag_history );
 831
 832                                 if ( !array_pop ( $has_opened_tr ) ) {
 833                                         $line = "<tr><td></td></tr>{$line}";
 834                                 }
 835
 836                                 if ( array_pop ( $tr_history ) ) {
 837                                         $line = "</tr>{$line}";
 838                                 }
 839
 840                                 if ( array_pop ( $td_history ) ) {
 841                                         $line = "</{$last_tag}>{$line}";
 842                                 }
 843                                 array_pop ( $tr_attributes );
 844                                 $lines[$key] = $line . str_repeat( '</dd></dl>' , $indent_level );
 845                         } else if ( substr ( $line , 0 , 2 ) == '|-' ) {
 846                                 // Now we have a table row
 847                                 $line = preg_replace( '#^\|-+#', '', $line );
 848
 849                                 // Whats after the tag is now only attributes
 850                                 $attributes = $this->mStripState->unstripBoth( $line );
 851                                 $attributes = Sanitizer::fixTagAttributes ( $attributes , 'tr' );
 852                                 array_pop ( $tr_attributes );
 853                                 array_push ( $tr_attributes , $attributes );
 854
 855                                 $line = '';
 856                                 $last_tag = array_pop ( $last_tag_history );
 857                                 array_pop ( $has_opened_tr );
 858                                 array_push ( $has_opened_tr , true );
 859
 860                                 if ( array_pop ( $tr_history ) ) {
 861                                         $line = '</tr>';
 862                                 }
 863
 864                                 if ( array_pop ( $td_history ) ) {
 865                                         $line = "</{$last_tag}>{$line}";
 866                                 }
 867
 868                                 $lines[$key] = $line;
 869                                 array_push ( $tr_history , false );
 870                                 array_push ( $td_history , false );
 871                                 array_push ( $last_tag_history , '' );
 872                         }
 873                         else if ( $first_character == '|' || $first_character == '!' || substr ( $line , 0 , 2 )  == '|+' ) {
 874                                 // This might be cell elements, td, th or captions
 875                                 if ( substr ( $line , 0 , 2 ) == '|+' ) {
 876                                         $first_character = '+';
 877                                         $line = substr ( $line , 1 );
 878                                 }
 879
 880                                 $line = substr ( $line , 1 );
 881
 882                                 if ( $first_character == '!' ) {
 883                                         $line = str_replace ( '!!' , '||' , $line );
 884                                 }
 885
 886                                 // Split up multiple cells on the same line.
 887                                 // FIXME : This can result in improper nesting of tags processed
 888                                 // by earlier parser steps, but should avoid splitting up eg
 889                                 // attribute values containing literal "||".
 890                                 $cells = StringUtils::explodeMarkup( '||' , $line );
 891
 892                                 $lines[$key] = '';
 893
 894                                 // Loop through each table cell
 895                                 foreach ( $cells as $cell )
 896                                 {
 897                                         $previous = '';
 898                                         if ( $first_character != '+' )
 899                                         {
 900                                                 $tr_after = array_pop ( $tr_attributes );
 901                                                 if ( !array_pop ( $tr_history ) ) {
 902                                                         $previous = "<tr{$tr_after}>\n";
 903                                                 }
 904                                                 array_push ( $tr_history , true );
 905                                                 array_push ( $tr_attributes , '' );
 906                                                 array_pop ( $has_opened_tr );
 907                                                 array_push ( $has_opened_tr , true );
 908                                         }
 909
 910                                         $last_tag = array_pop ( $last_tag_history );
 911
 912                                         if ( array_pop ( $td_history ) ) {
 913                                                 $previous = "</{$last_tag}>{$previous}";
 914                                         }
 915
 916                                         if ( $first_character == '|' ) {
 917                                                 $last_tag = 'td';
 918                                         } else if ( $first_character == '!' ) {
 919                                                 $last_tag = 'th';
 920                                         } else if ( $first_character == '+' ) {
 921                                                 $last_tag = 'caption';
 922                                         } else {
 923                                                 $last_tag = '';
 924                                         }
 925
 926                                         array_push ( $last_tag_history , $last_tag );
 927
 928                                         // A cell could contain both parameters and data
 929                                         $cell_data = explode ( '|' , $cell , 2 );
 930
 931                                         // Bug 553: Note that a '|' inside an invalid link should not
 932                                         // be mistaken as delimiting cell parameters
 933                                         if ( strpos( $cell_data[0], '[[' ) !== false ) {
 934                                                 $cell = "{$previous}<{$last_tag}>{$cell}";
 935                                         } else if ( count ( $cell_data ) == 1 )
 936                                                 $cell = "{$previous}<{$last_tag}>{$cell_data[0]}";
 937                                         else {
 938                                                 $attributes = $this->mStripState->unstripBoth( $cell_data[0] );
 939                                                 $attributes = Sanitizer::fixTagAttributes( $attributes , $last_tag );
 940                                                 $cell = "{$previous}<{$last_tag}{$attributes}>{$cell_data[1]}";
 941                                         }
 942
 943                                         $lines[$key] .= $cell;
 944                                         array_push ( $td_history , true );
 945                                 }
 946                         }
 947                 }
 948
 949                 // Closing open td, tr && table
 950                 while ( count ( $td_history ) > 0 )
 951                 {
 952                         if ( array_pop ( $td_history ) ) {
 953                                 $lines[] = '</td>' ;
 954                         }
 955                         if ( array_pop ( $tr_history ) ) {
 956                                 $lines[] = '</tr>' ;
 957                         }
 958                         if ( !array_pop ( $has_opened_tr ) ) {
 959                                 $lines[] = "<tr><td></td></tr>" ;
 960                         }
 961
 962                         $lines[] = '</table>' ;
 963                 }
 964
 965                 $output = implode ( "\n" , $lines ) ;
 966
 967                 // special case: don't return empty table
 968                 if( $output == "<table>\n<tr><td></td></tr>\n</table>" ) {
 969                         $output = '';
 970                 }
 971
 972                 wfProfileOut( $fname );
 973
 974                 return $output;
 975         }
 976
 977         /**
 978          * Helper function for parse() that transforms wiki markup into
 979          * HTML. Only called for $mOutputType == OT_HTML.
 980          *
 981          * @private
 982          */
 983         function internalParse( $text ) {
 984                 $args = array();
 985                 $isMain = true;
 986                 $fname = 'Parser::internalParse';
 987                 wfProfileIn( $fname );
 988
 989                 # Hook to suspend the parser in this state
 990                 if ( !wfRunHooks( 'ParserBeforeInternalParse', array( &$this, &$text, &$this->mStripState ) ) ) {
 991                         wfProfileOut( $fname );
 992                         return $text ;
 993                 }
 994
 995                 # Remove <noinclude> tags and <includeonly> sections
 996                 $text = strtr( $text, array( '<onlyinclude>' => '' , '</onlyinclude>' => '' ) );
 997                 $text = strtr( $text, array( '<noinclude>' => '', '</noinclude>' => '') );
 998                 $text = StringUtils::delimiterReplace( '<includeonly>', '</includeonly>', '', $text );
 999
1000                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'attributeStripCallback' ) );
1001
1002                 $text = $this->replaceVariables( $text, $args );
1003
1004                 // Tables need to come after variable replacement for things to work
1005                 // properly; putting them before other transformations should keep
1006                 // exciting things like link expansions from showing up in surprising
1007                 // places.
1008                 $text = $this->doTableStuff( $text );
1009
1010                 $text = preg_replace( '/(^|\n)-----*/', '\\1<hr />', $text );
1011
1012                 $text = $this->stripToc( $text );
1013                 $this->stripNoGallery( $text );
1014                 $text = $this->doHeadings( $text );
1015                 if($this->mOptions->getUseDynamicDates()) {
1016                         $df =& DateFormatter::getInstance();
1017                         $text = $df->reformat( $this->mOptions->getDateFormat(), $text );
1018                 }
1019                 $text = $this->doAllQuotes( $text );
1020                 $text = $this->replaceInternalLinks( $text );
1021                 $text = $this->replaceExternalLinks( $text );
1022
1023                 # replaceInternalLinks may sometimes leave behind
1024                 # absolute URLs, which have to be masked to hide them from replaceExternalLinks
1025                 $text = str_replace($this->mUniqPrefix."NOPARSE", "", $text);
1026
1027                 $text = $this->doMagicLinks( $text );
1028                 $text = $this->formatHeadings( $text, $isMain );
1029
1030                 wfProfileOut( $fname );
1031                 return $text;
1032         }
1033
1034         /**
1035          * Replace special strings like "ISBN xxx" and "RFC xxx" with
1036          * magic external links.
1037          *
1038          * @private
1039          */
1040         function &doMagicLinks( &$text ) {
1041                 wfProfileIn( __METHOD__ );
1042                 $text = preg_replace_callback(
1043                         '!(?:                           # Start cases
1044                             <a.*?</a> |                 # Skip link text
1045                             <.*?> |                     # Skip stuff inside HTML elements
1046                             (?:RFC|PMID)\s+([0-9]+) |   # RFC or PMID, capture number as m[1]
1047                             ISBN\s+(\b                  # ISBN, capture number as m[2]
1048                                      (?: 97[89] [\ \-]? )?   # optional 13-digit ISBN prefix
1049                                      (?: [0-9]  [\ \-]? ){9} # 9 digits with opt. delimiters
1050                                      [0-9Xx]                 # check digit
1051                                    \b)
1052                         )!x', array( &$this, 'magicLinkCallback' ), $text );
1053                 wfProfileOut( __METHOD__ );
1054                 return $text;
1055         }
1056
1057         function magicLinkCallback( $m ) {
1058                 if ( substr( $m[0], 0, 1 ) == '<' ) {
1059                         # Skip HTML element
1060                         return $m[0];
1061                 } elseif ( substr( $m[0], 0, 4 ) == 'ISBN' ) {
1062                         $isbn = $m[2];
1063                         $num = strtr( $isbn, array(
1064                                 '-' => '',
1065                                 ' ' => '',
1066                                 'x' => 'X',
1067                         ));
1068                         $titleObj = SpecialPage::getTitleFor( 'Booksources' );
1069                         $text = '<a href="' .
1070                                 $titleObj->escapeLocalUrl( "isbn=$num" ) .
1071                                 "\" class=\"internal\">ISBN $isbn</a>";
1072                 } else {
1073                         if ( substr( $m[0], 0, 3 ) == 'RFC' ) {
1074                                 $keyword = 'RFC';
1075                                 $urlmsg = 'rfcurl';
1076                                 $id = $m[1];
1077                         } elseif ( substr( $m[0], 0, 4 ) == 'PMID' ) {
1078                                 $keyword = 'PMID';
1079                                 $urlmsg = 'pubmedurl';
1080                                 $id = $m[1];
1081                         } else {
1082                                 throw new MWException( __METHOD__.': unrecognised match type "' .
1083                                         substr($m[0], 0, 20 ) . '"' );
1084                         }
1085
1086                         $url = wfMsg( $urlmsg, $id);
1087                         $sk = $this->mOptions->getSkin();
1088                         $la = $sk->getExternalLinkAttributes( $url, $keyword.$id );
1089                         $text = "<a href=\"{$url}\"{$la}>{$keyword} {$id}</a>";
1090                 }
1091                 return $text;
1092         }
1093
1094         /**
1095          * Parse headers and return html
1096          *
1097          * @private
1098          */
1099         function doHeadings( $text ) {
1100                 $fname = 'Parser::doHeadings';
1101                 wfProfileIn( $fname );
1102                 for ( $i = 6; $i >= 1; --$i ) {
1103                         $h = str_repeat( '=', $i );
1104                         $text = preg_replace( "/^{$h}(.+){$h}\\s*$/m",
1105                           "<h{$i}>\\1</h{$i}>\\2", $text );
1106                 }
1107                 wfProfileOut( $fname );
1108                 return $text;
1109         }
1110
1111         /**
1112          * Replace single quotes with HTML markup
1113          * @private
1114          * @return string the altered text
1115          */
1116         function doAllQuotes( $text ) {
1117                 $fname = 'Parser::doAllQuotes';
1118                 wfProfileIn( $fname );
1119                 $outtext = '';
1120                 $lines = explode( "\n", $text );
1121                 foreach ( $lines as $line ) {
1122                         $outtext .= $this->doQuotes ( $line ) . "\n";
1123                 }
1124                 $outtext = substr($outtext, 0,-1);
1125                 wfProfileOut( $fname );
1126                 return $outtext;
1127         }
1128
1129         /**
1130          * Helper function for doAllQuotes()
1131          * @private
1132          */
1133         function doQuotes( $text ) {
1134                 $arr = preg_split( "/(''+)/", $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1135                 if ( count( $arr ) == 1 )
1136                         return $text;
1137                 else
1138                 {
1139                         # First, do some preliminary work. This may shift some apostrophes from
1140                         # being mark-up to being text. It also counts the number of occurrences
1141                         # of bold and italics mark-ups.
1142                         $i = 0;
1143                         $numbold = 0;
1144                         $numitalics = 0;
1145                         foreach ( $arr as $r )
1146                         {
1147                                 if ( ( $i % 2 ) == 1 )
1148                                 {
1149                                         # If there are ever four apostrophes, assume the first is supposed to
1150                                         # be text, and the remaining three constitute mark-up for bold text.
1151                                         if ( strlen( $arr[$i] ) == 4 )
1152                                         {
1153                                                 $arr[$i-1] .= "'";
1154                                                 $arr[$i] = "'''";
1155                                         }
1156                                         # If there are more than 5 apostrophes in a row, assume they're all
1157                                         # text except for the last 5.
1158                                         else if ( strlen( $arr[$i] ) > 5 )
1159                                         {
1160                                                 $arr[$i-1] .= str_repeat( "'", strlen( $arr[$i] ) - 5 );
1161                                                 $arr[$i] = "'''''";
1162                                         }
1163                                         # Count the number of occurrences of bold and italics mark-ups.
1164                                         # We are not counting sequences of five apostrophes.
1165                                         if ( strlen( $arr[$i] ) == 2 )      { $numitalics++;             }
1166                                         else if ( strlen( $arr[$i] ) == 3 ) { $numbold++;                }
1167                                         else if ( strlen( $arr[$i] ) == 5 ) { $numitalics++; $numbold++; }
1168                                 }
1169                                 $i++;
1170                         }
1171
1172                         # If there is an odd number of both bold and italics, it is likely
1173                         # that one of the bold ones was meant to be an apostrophe followed
1174                         # by italics. Which one we cannot know for certain, but it is more
1175                         # likely to be one that has a single-letter word before it.
1176                         if ( ( $numbold % 2 == 1 ) && ( $numitalics % 2 == 1 ) )
1177                         {
1178                                 $i = 0;
1179                                 $firstsingleletterword = -1;
1180                                 $firstmultiletterword = -1;
1181                                 $firstspace = -1;
1182                                 foreach ( $arr as $r )
1183                                 {
1184                                         if ( ( $i % 2 == 1 ) and ( strlen( $r ) == 3 ) )
1185                                         {
1186                                                 $x1 = substr ($arr[$i-1], -1);
1187                                                 $x2 = substr ($arr[$i-1], -2, 1);
1188                                                 if ($x1 == ' ') {
1189                                                         if ($firstspace == -1) $firstspace = $i;
1190                                                 } else if ($x2 == ' ') {
1191                                                         if ($firstsingleletterword == -1) $firstsingleletterword = $i;
1192                                                 } else {
1193                                                         if ($firstmultiletterword == -1) $firstmultiletterword = $i;
1194                                                 }
1195                                         }
1196                                         $i++;
1197                                 }
1198
1199                                 # If there is a single-letter word, use it!
1200                                 if ($firstsingleletterword > -1)
1201                                 {
1202                                         $arr [ $firstsingleletterword ] = "''";
1203                                         $arr [ $firstsingleletterword-1 ] .= "'";
1204                                 }
1205                                 # If not, but there's a multi-letter word, use that one.
1206                                 else if ($firstmultiletterword > -1)
1207                                 {
1208                                         $arr [ $firstmultiletterword ] = "''";
1209                                         $arr [ $firstmultiletterword-1 ] .= "'";
1210                                 }
1211                                 # ... otherwise use the first one that has neither.
1212                                 # (notice that it is possible for all three to be -1 if, for example,
1213                                 # there is only one pentuple-apostrophe in the line)
1214                                 else if ($firstspace > -1)
1215                                 {
1216                                         $arr [ $firstspace ] = "''";
1217                                         $arr [ $firstspace-1 ] .= "'";
1218                                 }
1219                         }
1220
1221                         # Now let's actually convert our apostrophic mush to HTML!
1222                         $output = '';
1223                         $buffer = '';
1224                         $state = '';
1225                         $i = 0;
1226                         foreach ($arr as $r)
1227                         {
1228                                 if (($i % 2) == 0)
1229                                 {
1230                                         if ($state == 'both')
1231                                                 $buffer .= $r;
1232                                         else
1233                                                 $output .= $r;
1234                                 }
1235                                 else
1236                                 {
1237                                         if (strlen ($r) == 2)
1238                                         {
1239                                                 if ($state == 'i')
1240                                                 { $output .= '</i>'; $state = ''; }
1241                                                 else if ($state == 'bi')
1242                                                 { $output .= '</i>'; $state = 'b'; }
1243                                                 else if ($state == 'ib')
1244                                                 { $output .= '</b></i><b>'; $state = 'b'; }
1245                                                 else if ($state == 'both')
1246                                                 { $output .= '<b><i>'.$buffer.'</i>'; $state = 'b'; }
1247                                                 else # $state can be 'b' or ''
1248                                                 { $output .= '<i>'; $state .= 'i'; }
1249                                         }
1250                                         else if (strlen ($r) == 3)
1251                                         {
1252                                                 if ($state == 'b')
1253                                                 { $output .= '</b>'; $state = ''; }
1254                                                 else if ($state == 'bi')
1255                                                 { $output .= '</i></b><i>'; $state = 'i'; }
1256                                                 else if ($state == 'ib')
1257                                                 { $output .= '</b>'; $state = 'i'; }
1258                                                 else if ($state == 'both')
1259                                                 { $output .= '<i><b>'.$buffer.'</b>'; $state = 'i'; }
1260                                                 else # $state can be 'i' or ''
1261                                                 { $output .= '<b>'; $state .= 'b'; }
1262                                         }
1263                                         else if (strlen ($r) == 5)
1264                                         {
1265                                                 if ($state == 'b')
1266                                                 { $output .= '</b><i>'; $state = 'i'; }
1267                                                 else if ($state == 'i')
1268                                                 { $output .= '</i><b>'; $state = 'b'; }
1269                                                 else if ($state == 'bi')
1270                                                 { $output .= '</i></b>'; $state = ''; }
1271                                                 else if ($state == 'ib')
1272                                                 { $output .= '</b></i>'; $state = ''; }
1273                                                 else if ($state == 'both')
1274                                                 { $output .= '<i><b>'.$buffer.'</b></i>'; $state = ''; }
1275                                                 else # ($state == '')
1276                                                 { $buffer = ''; $state = 'both'; }
1277                                         }
1278                                 }
1279                                 $i++;
1280                         }
1281                         # Now close all remaining tags.  Notice that the order is important.
1282                         if ($state == 'b' || $state == 'ib')
1283                                 $output .= '</b>';
1284                         if ($state == 'i' || $state == 'bi' || $state == 'ib')
1285                                 $output .= '</i>';
1286                         if ($state == 'bi')
1287                                 $output .= '</b>';
1288                         # There might be lonely ''''', so make sure we have a buffer
1289                         if ($state == 'both' && $buffer)
1290                                 $output .= '<b><i>'.$buffer.'</i></b>';
1291                         return $output;
1292                 }
1293         }
1294
1295         /**
1296          * Replace external links
1297          *
1298          * Note: this is all very hackish and the order of execution matters a lot.
1299          * Make sure to run maintenance/parserTests.php if you change this code.
1300          *
1301          * @private
1302          */
1303         function replaceExternalLinks( $text ) {
1304                 global $wgContLang;
1305                 $fname = 'Parser::replaceExternalLinks';
1306                 wfProfileIn( $fname );
1307
1308                 $sk = $this->mOptions->getSkin();
1309
1310                 $bits = preg_split( EXT_LINK_BRACKETED, $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1311
1312                 $s = $this->replaceFreeExternalLinks( array_shift( $bits ) );
1313
1314                 $i = 0;
1315                 while ( $i<count( $bits ) ) {
1316                         $url = $bits[$i++];
1317                         $protocol = $bits[$i++];
1318                         $text = $bits[$i++];
1319                         $trail = $bits[$i++];
1320
1321                         # The characters '<' and '>' (which were escaped by
1322                         # removeHTMLtags()) should not be included in
1323                         # URLs, per RFC 2396.
1324                         $m2 = array();
1325                         if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1326                                 $text = substr($url, $m2[0][1]) . ' ' . $text;
1327                                 $url = substr($url, 0, $m2[0][1]);
1328                         }
1329
1330                         # If the link text is an image URL, replace it with an <img> tag
1331                         # This happened by accident in the original parser, but some people used it extensively
1332                         $img = $this->maybeMakeExternalImage( $text );
1333                         if ( $img !== false ) {
1334                                 $text = $img;
1335                         }
1336
1337                         $dtrail = '';
1338
1339                         # Set linktype for CSS - if URL==text, link is essentially free
1340                         $linktype = ($text == $url) ? 'free' : 'text';
1341
1342                         # No link text, e.g. [http://domain.tld/some.link]
1343                         if ( $text == '' ) {
1344                                 # Autonumber if allowed. See bug #5918
1345                                 if ( strpos( wfUrlProtocols(), substr($protocol, 0, strpos($protocol, ':')) ) !== false ) {
1346                                         $text = '[' . ++$this->mAutonumber . ']';
1347                                         $linktype = 'autonumber';
1348                                 } else {
1349                                         # Otherwise just use the URL
1350                                         $text = htmlspecialchars( $url );
1351                                         $linktype = 'free';
1352                                 }
1353                         } else {
1354                                 # Have link text, e.g. [http://domain.tld/some.link text]s
1355                                 # Check for trail
1356                                 list( $dtrail, $trail ) = Linker::splitTrail( $trail );
1357                         }
1358
1359                         $text = $wgContLang->markNoConversion($text);
1360
1361                         $url = Sanitizer::cleanUrl( $url );
1362
1363                         # Process the trail (i.e. everything after this link up until start of the next link),
1364                         # replacing any non-bracketed links
1365                         $trail = $this->replaceFreeExternalLinks( $trail );
1366
1367                         # Use the encoded URL
1368                         # This means that users can paste URLs directly into the text
1369                         # Funny characters like &ouml; aren't valid in URLs anyway
1370                         # This was changed in August 2004
1371                         $s .= $sk->makeExternalLink( $url, $text, false, $linktype, $this->mTitle->getNamespace() ) . $dtrail . $trail;
1372
1373                         # Register link in the output object.
1374                         # Replace unnecessary URL escape codes with the referenced character
1375                         # This prevents spammers from hiding links from the filters
1376                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1377                         $this->mOutput->addExternalLink( $pasteurized );
1378                 }
1379
1380                 wfProfileOut( $fname );
1381                 return $s;
1382         }
1383
1384         /**
1385          * Replace anything that looks like a URL with a link
1386          * @private
1387          */
1388         function replaceFreeExternalLinks( $text ) {
1389                 global $wgContLang;
1390                 $fname = 'Parser::replaceFreeExternalLinks';
1391                 wfProfileIn( $fname );
1392
1393                 $bits = preg_split( '/(\b(?:' . wfUrlProtocols() . '))/S', $text, -1, PREG_SPLIT_DELIM_CAPTURE );
1394                 $s = array_shift( $bits );
1395                 $i = 0;
1396
1397                 $sk = $this->mOptions->getSkin();
1398
1399                 while ( $i < count( $bits ) ){
1400                         $protocol = $bits[$i++];
1401                         $remainder = $bits[$i++];
1402
1403                         $m = array();
1404                         if ( preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $remainder, $m ) ) {
1405                                 # Found some characters after the protocol that look promising
1406                                 $url = $protocol . $m[1];
1407                                 $trail = $m[2];
1408
1409                                 # special case: handle urls as url args:
1410                                 # http://www.example.com/foo?=http://www.example.com/bar
1411                                 if(strlen($trail) == 0 &&
1412                                         isset($bits[$i]) &&
1413                                         preg_match('/^'. wfUrlProtocols() . '$/S', $bits[$i]) &&
1414                                         preg_match( '/^('.EXT_LINK_URL_CLASS.'+)(.*)$/s', $bits[$i + 1], $m ))
1415                                 {
1416                                         # add protocol, arg
1417                                         $url .= $bits[$i] . $m[1]; # protocol, url as arg to previous link
1418                                         $i += 2;
1419                                         $trail = $m[2];
1420                                 }
1421
1422                                 # The characters '<' and '>' (which were escaped by
1423                                 # removeHTMLtags()) should not be included in
1424                                 # URLs, per RFC 2396.
1425                                 $m2 = array();
1426                                 if (preg_match('/&(lt|gt);/', $url, $m2, PREG_OFFSET_CAPTURE)) {
1427                                         $trail = substr($url, $m2[0][1]) . $trail;
1428                                         $url = substr($url, 0, $m2[0][1]);
1429                                 }
1430
1431                                 # Move trailing punctuation to $trail
1432                                 $sep = ',;\.:!?';
1433                                 # If there is no left bracket, then consider right brackets fair game too
1434                                 if ( strpos( $url, '(' ) === false ) {
1435                                         $sep .= ')';
1436                                 }
1437
1438                                 $numSepChars = strspn( strrev( $url ), $sep );
1439                                 if ( $numSepChars ) {
1440                                         $trail = substr( $url, -$numSepChars ) . $trail;
1441                                         $url = substr( $url, 0, -$numSepChars );
1442                                 }
1443
1444                                 $url = Sanitizer::cleanUrl( $url );
1445
1446                                 # Is this an external image?
1447                                 $text = $this->maybeMakeExternalImage( $url );
1448                                 if ( $text === false ) {
1449                                         # Not an image, make a link
1450                                         $text = $sk->makeExternalLink( $url, $wgContLang->markNoConversion($url), true, 'free', $this->mTitle->getNamespace() );
1451                                         # Register it in the output object...
1452                                         # Replace unnecessary URL escape codes with their equivalent characters
1453                                         $pasteurized = Parser::replaceUnusualEscapes( $url );
1454                                         $this->mOutput->addExternalLink( $pasteurized );
1455                                 }
1456                                 $s .= $text . $trail;
1457                         } else {
1458                                 $s .= $protocol . $remainder;
1459                         }
1460                 }
1461                 wfProfileOut( $fname );
1462                 return $s;
1463         }
1464
1465         /**
1466          * Replace unusual URL escape codes with their equivalent characters
1467          * @param string
1468          * @return string
1469          * @static
1470          * @fixme This can merge genuinely required bits in the path or query string,
1471          *        breaking legit URLs. A proper fix would treat the various parts of
1472          *        the URL differently; as a workaround, just use the output for
1473          *        statistical records, not for actual linking/output.
1474          */
1475         static function replaceUnusualEscapes( $url ) {
1476                 return preg_replace_callback( '/%[0-9A-Fa-f]{2}/',
1477                         array( 'Parser', 'replaceUnusualEscapesCallback' ), $url );
1478         }
1479
1480         /**
1481          * Callback function used in replaceUnusualEscapes().
1482          * Replaces unusual URL escape codes with their equivalent character
1483          * @static
1484          * @private
1485          */
1486         private static function replaceUnusualEscapesCallback( $matches ) {
1487                 $char = urldecode( $matches[0] );
1488                 $ord = ord( $char );
1489                 // Is it an unsafe or HTTP reserved character according to RFC 1738?
1490                 if ( $ord > 32 && $ord < 127 && strpos( '<>"#{}|\^~[]`;/?', $char ) === false ) {
1491                         // No, shouldn't be escaped
1492                         return $char;
1493                 } else {
1494                         // Yes, leave it escaped
1495                         return $matches[0];
1496                 }
1497         }
1498
1499         /**
1500          * make an image if it's allowed, either through the global
1501          * option or through the exception
1502          * @private
1503          */
1504         function maybeMakeExternalImage( $url ) {
1505                 $sk = $this->mOptions->getSkin();
1506                 $imagesfrom = $this->mOptions->getAllowExternalImagesFrom();
1507                 $imagesexception = !empty($imagesfrom);
1508                 $text = false;
1509                 if ( $this->mOptions->getAllowExternalImages()
1510                      || ( $imagesexception && strpos( $url, $imagesfrom ) === 0 ) ) {
1511                         if ( preg_match( EXT_IMAGE_REGEX, $url ) ) {
1512                                 # Image found
1513                                 $text = $sk->makeExternalImage( htmlspecialchars( $url ) );
1514                         }
1515                 }
1516                 return $text;
1517         }
1518
1519         /**
1520          * Process [[ ]] wikilinks
1521          *
1522          * @private
1523          */
1524         function replaceInternalLinks( $s ) {
1525                 global $wgContLang;
1526                 static $fname = 'Parser::replaceInternalLinks' ;
1527
1528                 wfProfileIn( $fname );
1529
1530                 wfProfileIn( $fname.'-setup' );
1531                 static $tc = FALSE;
1532                 # the % is needed to support urlencoded titles as well
1533                 if ( !$tc ) { $tc = Title::legalChars() . '#%'; }
1534
1535                 $sk = $this->mOptions->getSkin();
1536
1537                 #split the entire text string on occurences of [[
1538                 $a = explode( '[[', ' ' . $s );
1539                 #get the first element (all text up to first [[), and remove the space we added
1540                 $s = array_shift( $a );
1541                 $s = substr( $s, 1 );
1542
1543                 # Match a link having the form [[namespace:link|alternate]]trail
1544                 static $e1 = FALSE;
1545                 if ( !$e1 ) { $e1 = "/^([{$tc}]+)(?:\\|(.+?))?]](.*)\$/sD"; }
1546                 # Match cases where there is no "]]", which might still be images
1547                 static $e1_img = FALSE;
1548                 if ( !$e1_img ) { $e1_img = "/^([{$tc}]+)\\|(.*)\$/sD"; }
1549                 # Match the end of a line for a word that's not followed by whitespace,
1550                 # e.g. in the case of 'The Arab al[[Razi]]', 'al' will be matched
1551                 $e2 = wfMsgForContent( 'linkprefix' );
1552
1553                 $useLinkPrefixExtension = $wgContLang->linkPrefixExtension();
1554                 if( is_null( $this->mTitle ) ) {
1555                         throw new MWException( __METHOD__.": \$this->mTitle is null\n" );
1556                 }
1557                 $nottalk = !$this->mTitle->isTalkPage();
1558
1559                 if ( $useLinkPrefixExtension ) {
1560                         $m = array();
1561                         if ( preg_match( $e2, $s, $m ) ) {
1562                                 $first_prefix = $m[2];
1563                         } else {
1564                                 $first_prefix = false;
1565                         }
1566                 } else {
1567                         $prefix = '';
1568                 }
1569
1570                 if($wgContLang->hasVariants()) {
1571                         $selflink = $wgContLang->convertLinkToAllVariants($this->mTitle->getPrefixedText());
1572                 } else {
1573                         $selflink = array($this->mTitle->getPrefixedText());
1574                 }
1575                 $useSubpages = $this->areSubpagesAllowed();
1576                 wfProfileOut( $fname.'-setup' );
1577
1578                 # Loop for each link
1579                 for ($k = 0; isset( $a[$k] ); $k++) {
1580                         $line = $a[$k];
1581                         if ( $useLinkPrefixExtension ) {
1582                                 wfProfileIn( $fname.'-prefixhandling' );
1583                                 if ( preg_match( $e2, $s, $m ) ) {
1584                                         $prefix = $m[2];
1585                                         $s = $m[1];
1586                                 } else {
1587                                         $prefix='';
1588                                 }
1589                                 # first link
1590                                 if($first_prefix) {
1591                                         $prefix = $first_prefix;
1592                                         $first_prefix = false;
1593                                 }
1594                                 wfProfileOut( $fname.'-prefixhandling' );
1595                         }
1596
1597                         $might_be_img = false;
1598
1599                         wfProfileIn( "$fname-e1" );
1600                         if ( preg_match( $e1, $line, $m ) ) { # page with normal text or alt
1601                                 $text = $m[2];
1602                                 # If we get a ] at the beginning of $m[3] that means we have a link that's something like:
1603                                 # [[Image:Foo.jpg|[http://example.com desc]]] <- having three ] in a row fucks up,
1604                                 # the real problem is with the $e1 regex
1605                                 # See bug 1300.
1606                                 #
1607                                 # Still some problems for cases where the ] is meant to be outside punctuation,
1608                                 # and no image is in sight. See bug 2095.
1609                                 #
1610                                 if( $text !== '' &&
1611                                         substr( $m[3], 0, 1 ) === ']' &&
1612                                         strpos($text, '[') !== false
1613                                 )
1614                                 {
1615                                         $text .= ']'; # so that replaceExternalLinks($text) works later
1616                                         $m[3] = substr( $m[3], 1 );
1617                                 }
1618                                 # fix up urlencoded title texts
1619                                 if( strpos( $m[1], '%' ) !== false ) {
1620                                         # Should anchors '#' also be rejected?
1621                                         $m[1] = str_replace( array('<', '>'), array('&lt;', '&gt;'), urldecode($m[1]) );
1622                                 }
1623                                 $trail = $m[3];
1624                         } elseif( preg_match($e1_img, $line, $m) ) { # Invalid, but might be an image with a link in its caption
1625                                 $might_be_img = true;
1626                                 $text = $m[2];
1627                                 if ( strpos( $m[1], '%' ) !== false ) {
1628                                        $m[1] = urldecode($m[1]);
1629                                 }
1630                                 $trail = "";
1631                         } else { # Invalid form; output directly
1632                                 $s .= $prefix . '[[' . $line ;
1633                                 wfProfileOut( "$fname-e1" );
1634                                 continue;
1635                         }
1636                         wfProfileOut( "$fname-e1" );
1637                         wfProfileIn( "$fname-misc" );
1638
1639                         # Don't allow internal links to pages containing
1640                         # PROTO: where PROTO is a valid URL protocol; these
1641                         # should be external links.
1642                         if (preg_match('/^(\b(?:' . wfUrlProtocols() . '))/', $m[1])) {
1643                                 $s .= $prefix . '[[' . $line ;
1644                                 continue;
1645                         }
1646
1647                         # Make subpage if necessary
1648                         if( $useSubpages ) {
1649                                 $link = $this->maybeDoSubpageLink( $m[1], $text );
1650                         } else {
1651                                 $link = $m[1];
1652                         }
1653
1654                         $noforce = (substr($m[1], 0, 1) != ':');
1655                         if (!$noforce) {
1656                                 # Strip off leading ':'
1657                                 $link = substr($link, 1);
1658                         }
1659
1660                         wfProfileOut( "$fname-misc" );
1661                         wfProfileIn( "$fname-title" );
1662                         $nt = Title::newFromText( $this->mStripState->unstripNoWiki($link) );
1663                         if( !$nt ) {
1664                                 $s .= $prefix . '[[' . $line;
1665                                 wfProfileOut( "$fname-title" );
1666                                 continue;
1667                         }
1668
1669                         $ns = $nt->getNamespace();
1670                         $iw = $nt->getInterWiki();
1671                         wfProfileOut( "$fname-title" );
1672
1673                         if ($might_be_img) { # if this is actually an invalid link
1674                                 wfProfileIn( "$fname-might_be_img" );
1675                                 if ($ns == NS_IMAGE && $noforce) { #but might be an image
1676                                         $found = false;
1677                                         while (isset ($a[$k+1]) ) {
1678                                                 #look at the next 'line' to see if we can close it there
1679                                                 $spliced = array_splice( $a, $k + 1, 1 );
1680                                                 $next_line = array_shift( $spliced );
1681                                                 $m = explode( ']]', $next_line, 3 );
1682                                                 if ( count( $m ) == 3 ) {
1683                                                         # the first ]] closes the inner link, the second the image
1684                                                         $found = true;
1685                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1686                                                         $trail = $m[2];
1687                                                         break;
1688                                                 } elseif ( count( $m ) == 2 ) {
1689                                                         #if there's exactly one ]] that's fine, we'll keep looking
1690                                                         $text .= "[[{$m[0]}]]{$m[1]}";
1691                                                 } else {
1692                                                         #if $next_line is invalid too, we need look no further
1693                                                         $text .= '[[' . $next_line;
1694                                                         break;
1695                                                 }
1696                                         }
1697                                         if ( !$found ) {
1698                                                 # we couldn't find the end of this imageLink, so output it raw
1699                                                 #but don't ignore what might be perfectly normal links in the text we've examined
1700                                                 $text = $this->replaceInternalLinks($text);
1701                                                 $s .= "{$prefix}[[$link|$text";
1702                                                 # note: no $trail, because without an end, there *is* no trail
1703                                                 wfProfileOut( "$fname-might_be_img" );
1704                                                 continue;
1705                                         }
1706                                 } else { #it's not an image, so output it raw
1707                                         $s .= "{$prefix}[[$link|$text";
1708                                         # note: no $trail, because without an end, there *is* no trail
1709                                         wfProfileOut( "$fname-might_be_img" );
1710                                         continue;
1711                                 }
1712                                 wfProfileOut( "$fname-might_be_img" );
1713                         }
1714
1715                         $wasblank = ( '' == $text );
1716                         if( $wasblank ) $text = $link;
1717
1718                         # Link not escaped by : , create the various objects
1719                         if( $noforce ) {
1720
1721                                 # Interwikis
1722                                 wfProfileIn( "$fname-interwiki" );
1723                                 if( $iw && $this->mOptions->getInterwikiMagic() && $nottalk && $wgContLang->getLanguageName( $iw ) ) {
1724                                         $this->mOutput->addLanguageLink( $nt->getFullText() );
1725                                         $s = rtrim($s . $prefix);
1726                                         $s .= trim($trail, "\n") == '' ? '': $prefix . $trail;
1727                                         wfProfileOut( "$fname-interwiki" );
1728                                         continue;
1729                                 }
1730                                 wfProfileOut( "$fname-interwiki" );
1731
1732                                 if ( $ns == NS_IMAGE ) {
1733                                         wfProfileIn( "$fname-image" );
1734                                         if ( !wfIsBadImage( $nt->getDBkey(), $this->mTitle ) ) {
1735                                                 # recursively parse links inside the image caption
1736                                                 # actually, this will parse them in any other parameters, too,
1737                                                 # but it might be hard to fix that, and it doesn't matter ATM
1738                                                 $text = $this->replaceExternalLinks($text);
1739                                                 $text = $this->replaceInternalLinks($text);
1740
1741                                                 # cloak any absolute URLs inside the image markup, so replaceExternalLinks() won't touch them
1742                                                 $s .= $prefix . $this->armorLinks( $this->makeImage( $nt, $text ) ) . $trail;
1743                                                 $this->mOutput->addImage( $nt->getDBkey() );
1744
1745                                                 wfProfileOut( "$fname-image" );
1746                                                 continue;
1747                                         } else {
1748                                                 # We still need to record the image's presence on the page
1749                                                 $this->mOutput->addImage( $nt->getDBkey() );
1750                                         }
1751                                         wfProfileOut( "$fname-image" );
1752
1753                                 }
1754
1755                                 if ( $ns == NS_CATEGORY ) {
1756                                         wfProfileIn( "$fname-category" );
1757                                         $s = rtrim($s . "\n"); # bug 87
1758
1759                                         if ( $wasblank ) {
1760                                                 $sortkey = $this->getDefaultSort();
1761                                         } else {
1762                                                 $sortkey = $text;
1763                                         }
1764                                         $sortkey = Sanitizer::decodeCharReferences( $sortkey );
1765                                         $sortkey = str_replace( "\n", '', $sortkey );
1766                                         $sortkey = $wgContLang->convertCategoryKey( $sortkey );
1767                                         $this->mOutput->addCategory( $nt->getDBkey(), $sortkey );
1768
1769                                         /**
1770                                          * Strip the whitespace Category links produce, see bug 87
1771                                          * @todo We might want to use trim($tmp, "\n") here.
1772                                          */
1773                                         $s .= trim($prefix . $trail, "\n") == '' ? '': $prefix . $trail;
1774
1775                                         wfProfileOut( "$fname-category" );
1776                                         continue;
1777                                 }
1778                         }
1779
1780                         # Self-link checking
1781                         if( $nt->getFragment() === '' ) {
1782                                 if( in_array( $nt->getPrefixedText(), $selflink, true ) ) {
1783                                         $s .= $prefix . $sk->makeSelfLinkObj( $nt, $text, '', $trail );
1784                                         continue;
1785                                 }
1786                         }
1787
1788                         # Special and Media are pseudo-namespaces; no pages actually exist in them
1789                         if( $ns == NS_MEDIA ) {
1790                                 $link = $sk->makeMediaLinkObj( $nt, $text );
1791                                 # Cloak with NOPARSE to avoid replacement in replaceExternalLinks
1792                                 $s .= $prefix . $this->armorLinks( $link ) . $trail;
1793                                 $this->mOutput->addImage( $nt->getDBkey() );
1794                                 continue;
1795                         } elseif( $ns == NS_SPECIAL ) {
1796                                 $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1797                                 continue;
1798                         } elseif( $ns == NS_IMAGE ) {
1799                                 $img = new Image( $nt );
1800                                 if( $img->exists() ) {
1801                                         // Force a blue link if the file exists; may be a remote
1802                                         // upload on the shared repository, and we want to see its
1803                                         // auto-generated page.
1804                                         $s .= $this->makeKnownLinkHolder( $nt, $text, '', $trail, $prefix );
1805                                         $this->mOutput->addLink( $nt );
1806                                         continue;
1807                                 }
1808                         }
1809                         $s .= $this->makeLinkHolder( $nt, $text, '', $trail, $prefix );
1810                 }
1811                 wfProfileOut( $fname );
1812                 return $s;
1813         }
1814
1815         /**
1816          * Make a link placeholder. The text returned can be later resolved to a real link with
1817          * replaceLinkHolders(). This is done for two reasons: firstly to avoid further
1818          * parsing of interwiki links, and secondly to allow all existence checks and
1819          * article length checks (for stub links) to be bundled into a single query.
1820          *
1821          */
1822         function makeLinkHolder( &$nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1823                 wfProfileIn( __METHOD__ );
1824                 if ( ! is_object($nt) ) {
1825                         # Fail gracefully
1826                         $retVal = "<!-- ERROR -->{$prefix}{$text}{$trail}";
1827                 } else {
1828                         # Separate the link trail from the rest of the link
1829                         list( $inside, $trail ) = Linker::splitTrail( $trail );
1830
1831                         if ( $nt->isExternal() ) {
1832                                 $nr = array_push( $this->mInterwikiLinkHolders['texts'], $prefix.$text.$inside );
1833                                 $this->mInterwikiLinkHolders['titles'][] = $nt;
1834                                 $retVal = '<!--IWLINK '. ($nr-1) ."-->{$trail}";
1835                         } else {
1836                                 $nr = array_push( $this->mLinkHolders['namespaces'], $nt->getNamespace() );
1837                                 $this->mLinkHolders['dbkeys'][] = $nt->getDBkey();
1838                                 $this->mLinkHolders['queries'][] = $query;
1839                                 $this->mLinkHolders['texts'][] = $prefix.$text.$inside;
1840                                 $this->mLinkHolders['titles'][] = $nt;
1841
1842                                 $retVal = '<!--LINK '. ($nr-1) ."-->{$trail}";
1843                         }
1844                 }
1845                 wfProfileOut( __METHOD__ );
1846                 return $retVal;
1847         }
1848
1849         /**
1850          * Render a forced-blue link inline; protect against double expansion of
1851          * URLs if we're in a mode that prepends full URL prefixes to internal links.
1852          * Since this little disaster has to split off the trail text to avoid
1853          * breaking URLs in the following text without breaking trails on the
1854          * wiki links, it's been made into a horrible function.
1855          *
1856          * @param Title $nt
1857          * @param string $text
1858          * @param string $query
1859          * @param string $trail
1860          * @param string $prefix
1861          * @return string HTML-wikitext mix oh yuck
1862          */
1863         function makeKnownLinkHolder( $nt, $text = '', $query = '', $trail = '', $prefix = '' ) {
1864                 list( $inside, $trail ) = Linker::splitTrail( $trail );
1865                 $sk = $this->mOptions->getSkin();
1866                 $link = $sk->makeKnownLinkObj( $nt, $text, $query, $inside, $prefix );
1867                 return $this->armorLinks( $link ) . $trail;
1868         }
1869
1870         /**
1871          * Insert a NOPARSE hacky thing into any inline links in a chunk that's
1872          * going to go through further parsing steps before inline URL expansion.
1873          *
1874          * In particular this is important when using action=render, which causes
1875          * full URLs to be included.
1876          *
1877          * Oh man I hate our multi-layer parser!
1878          *
1879          * @param string more-or-less HTML
1880          * @return string less-or-more HTML with NOPARSE bits
1881          */
1882         function armorLinks( $text ) {
1883                 return preg_replace( '/\b(' . wfUrlProtocols() . ')/',
1884                         "{$this->mUniqPrefix}NOPARSE$1", $text );
1885         }
1886
1887         /**
1888          * Return true if subpage links should be expanded on this page.
1889          * @return bool
1890          */
1891         function areSubpagesAllowed() {
1892                 # Some namespaces don't allow subpages
1893                 global $wgNamespacesWithSubpages;
1894                 return !empty($wgNamespacesWithSubpages[$this->mTitle->getNamespace()]);
1895         }
1896
1897         /**
1898          * Handle link to subpage if necessary
1899          * @param string $target the source of the link
1900          * @param string &$text the link text, modified as necessary
1901          * @return string the full name of the link
1902          * @private
1903          */
1904         function maybeDoSubpageLink($target, &$text) {
1905                 # Valid link forms:
1906                 # Foobar -- normal
1907                 # :Foobar -- override special treatment of prefix (images, language links)
1908                 # /Foobar -- convert to CurrentPage/Foobar
1909                 # /Foobar/ -- convert to CurrentPage/Foobar, strip the initial / from text
1910                 # ../ -- convert to CurrentPage, from CurrentPage/CurrentSubPage
1911                 # ../Foobar -- convert to CurrentPage/Foobar, from CurrentPage/CurrentSubPage
1912
1913                 $fname = 'Parser::maybeDoSubpageLink';
1914                 wfProfileIn( $fname );
1915                 $ret = $target; # default return value is no change
1916
1917                 # bug 7425
1918                 $target = trim( $target );
1919
1920                 # Some namespaces don't allow subpages,
1921                 # so only perform processing if subpages are allowed
1922                 if( $this->areSubpagesAllowed() ) {
1923                         # Look at the first character
1924                         if( $target != '' && $target{0} == '/' ) {
1925                                 # / at end means we don't want the slash to be shown
1926                                 $trailingSlashes = preg_match_all( '%(/+)$%', $target, $m );
1927                                 if( $trailingSlashes ) {
1928                                         $noslash = $target = substr( $target, 1, -strlen($m[0][0]) );
1929                                 } else {
1930                                         $noslash = substr( $target, 1 );
1931                                 }
1932
1933                                 $ret = $this->mTitle->getPrefixedText(). '/' . trim($noslash);
1934                                 if( '' === $text ) {
1935                                         $text = $target;
1936                                 } # this might be changed for ugliness reasons
1937                         } else {
1938                                 # check for .. subpage backlinks
1939                                 $dotdotcount = 0;
1940                                 $nodotdot = $target;
1941                                 while( strncmp( $nodotdot, "../", 3 ) == 0 ) {
1942                                         ++$dotdotcount;
1943                                         $nodotdot = substr( $nodotdot, 3 );
1944                                 }
1945                                 if($dotdotcount > 0) {
1946                                         $exploded = explode( '/', $this->mTitle->GetPrefixedText() );
1947                                         if( count( $exploded ) > $dotdotcount ) { # not allowed to go below top level page
1948                                                 $ret = implode( '/', array_slice( $exploded, 0, -$dotdotcount ) );
1949                                                 # / at the end means don't show full path
1950                                                 if( substr( $nodotdot, -1, 1 ) == '/' ) {
1951                                                         $nodotdot = substr( $nodotdot, 0, -1 );
1952                                                         if( '' === $text ) {
1953                                                                 $text = $nodotdot;
1954                                                         }
1955                                                 }
1956                                                 $nodotdot = trim( $nodotdot );
1957                                                 if( $nodotdot != '' ) {
1958                                                         $ret .= '/' . $nodotdot;
1959                                                 }
1960                                         }
1961                                 }
1962                         }
1963                 }
1964
1965                 wfProfileOut( $fname );
1966                 return $ret;
1967         }
1968
1969         /**#@+
1970          * Used by doBlockLevels()
1971          * @private
1972          */
1973         /* private */ function closeParagraph() {
1974                 $result = '';
1975                 if ( '' != $this->mLastSection ) {
1976                         $result = '</' . $this->mLastSection  . ">\n";
1977                 }
1978                 $this->mInPre = false;
1979                 $this->mLastSection = '';
1980                 return $result;
1981         }
1982         # getCommon() returns the length of the longest common substring
1983         # of both arguments, starting at the beginning of both.
1984         #
1985         /* private */ function getCommon( $st1, $st2 ) {
1986                 $fl = strlen( $st1 );
1987                 $shorter = strlen( $st2 );
1988                 if ( $fl < $shorter ) { $shorter = $fl; }
1989
1990                 for ( $i = 0; $i < $shorter; ++$i ) {
1991                         if ( $st1{$i} != $st2{$i} ) { break; }
1992                 }
1993                 return $i;
1994         }
1995         # These next three functions open, continue, and close the list
1996         # element appropriate to the prefix character passed into them.
1997         #
1998         /* private */ function openList( $char ) {
1999                 $result = $this->closeParagraph();
2000
2001                 if ( '*' == $char ) { $result .= '<ul><li>'; }
2002                 else if ( '#' == $char ) { $result .= '<ol><li>'; }
2003                 else if ( ':' == $char ) { $result .= '<dl><dd>'; }
2004                 else if ( ';' == $char ) {
2005                         $result .= '<dl><dt>';
2006                         $this->mDTopen = true;
2007                 }
2008                 else { $result = '<!-- ERR 1 -->'; }
2009
2010                 return $result;
2011         }
2012
2013         /* private */ function nextItem( $char ) {
2014                 if ( '*' == $char || '#' == $char ) { return '</li><li>'; }
2015                 else if ( ':' == $char || ';' == $char ) {
2016                         $close = '</dd>';
2017                         if ( $this->mDTopen ) { $close = '</dt>'; }
2018                         if ( ';' == $char ) {
2019                                 $this->mDTopen = true;
2020                                 return $close . '<dt>';
2021                         } else {
2022                                 $this->mDTopen = false;
2023                                 return $close . '<dd>';
2024                         }
2025                 }
2026                 return '<!-- ERR 2 -->';
2027         }
2028
2029         /* private */ function closeList( $char ) {
2030                 if ( '*' == $char ) { $text = '</li></ul>'; }
2031                 else if ( '#' == $char ) { $text = '</li></ol>'; }
2032                 else if ( ':' == $char ) {
2033                         if ( $this->mDTopen ) {
2034                                 $this->mDTopen = false;
2035                                 $text = '</dt></dl>';
2036                         } else {
2037                                 $text = '</dd></dl>';
2038                         }
2039                 }
2040                 else {  return '<!-- ERR 3 -->'; }
2041                 return $text."\n";
2042         }
2043         /**#@-*/
2044
2045         /**
2046          * Make lists from lines starting with ':', '*', '#', etc.
2047          *
2048          * @private
2049          * @return string the lists rendered as HTML
2050          */
2051         function doBlockLevels( $text, $linestart ) {
2052                 $fname = 'Parser::doBlockLevels';
2053                 wfProfileIn( $fname );
2054
2055                 # Parsing through the text line by line.  The main thing
2056                 # happening here is handling of block-level elements p, pre,
2057                 # and making lists from lines starting with * # : etc.
2058                 #
2059                 $textLines = explode( "\n", $text );
2060
2061                 $lastPrefix = $output = '';
2062                 $this->mDTopen = $inBlockElem = false;
2063                 $prefixLength = 0;
2064                 $paragraphStack = false;
2065
2066                 if ( !$linestart ) {
2067                         $output .= array_shift( $textLines );
2068                 }
2069                 foreach ( $textLines as $oLine ) {
2070                         $lastPrefixLength = strlen( $lastPrefix );
2071                         $preCloseMatch = preg_match('/<\\/pre/i', $oLine );
2072                         $preOpenMatch = preg_match('/<pre/i', $oLine );
2073                         if ( !$this->mInPre ) {
2074                                 # Multiple prefixes may abut each other for nested lists.
2075                                 $prefixLength = strspn( $oLine, '*#:;' );
2076                                 $pref = substr( $oLine, 0, $prefixLength );
2077
2078                                 # eh?
2079                                 $pref2 = str_replace( ';', ':', $pref );
2080                                 $t = substr( $oLine, $prefixLength );
2081                                 $this->mInPre = !empty($preOpenMatch);
2082                         } else {
2083                                 # Don't interpret any other prefixes in preformatted text
2084                                 $prefixLength = 0;
2085                                 $pref = $pref2 = '';
2086                                 $t = $oLine;
2087                         }
2088
2089                         # List generation
2090                         if( $prefixLength && 0 == strcmp( $lastPrefix, $pref2 ) ) {
2091                                 # Same as the last item, so no need to deal with nesting or opening stuff
2092                                 $output .= $this->nextItem( substr( $pref, -1 ) );
2093                                 $paragraphStack = false;
2094
2095                                 if ( substr( $pref, -1 ) == ';') {
2096                                         # The one nasty exception: definition lists work like this:
2097                                         # ; title : definition text
2098                                         # So we check for : in the remainder text to split up the
2099                                         # title and definition, without b0rking links.
2100                                         $term = $t2 = '';
2101                                         if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2102                                                 $t = $t2;
2103                                                 $output .= $term . $this->nextItem( ':' );
2104                                         }
2105                                 }
2106                         } elseif( $prefixLength || $lastPrefixLength ) {
2107                                 # Either open or close a level...
2108                                 $commonPrefixLength = $this->getCommon( $pref, $lastPrefix );
2109                                 $paragraphStack = false;
2110
2111                                 while( $commonPrefixLength < $lastPrefixLength ) {
2112                                         $output .= $this->closeList( $lastPrefix{$lastPrefixLength-1} );
2113                                         --$lastPrefixLength;
2114                                 }
2115                                 if ( $prefixLength <= $commonPrefixLength && $commonPrefixLength > 0 ) {
2116                                         $output .= $this->nextItem( $pref{$commonPrefixLength-1} );
2117                                 }
2118                                 while ( $prefixLength > $commonPrefixLength ) {
2119                                         $char = substr( $pref, $commonPrefixLength, 1 );
2120                                         $output .= $this->openList( $char );
2121
2122                                         if ( ';' == $char ) {
2123                                                 # FIXME: This is dupe of code above
2124                                                 if ($this->findColonNoLinks($t, $term, $t2) !== false) {
2125                                                         $t = $t2;
2126                                                         $output .= $term . $this->nextItem( ':' );
2127                                                 }
2128                                         }
2129                                         ++$commonPrefixLength;
2130                                 }
2131                                 $lastPrefix = $pref2;
2132                         }
2133                         if( 0 == $prefixLength ) {
2134                                 wfProfileIn( "$fname-paragraph" );
2135                                 # No prefix (not in list)--go to paragraph mode
2136                                 // XXX: use a stack for nestable elements like span, table and div
2137                                 $openmatch = preg_match('/(<table|<blockquote|<h1|<h2|<h3|<h4|<h5|<h6|<pre|<tr|<p|<ul|<ol|<li|<\\/tr|<\\/td|<\\/th)/iS', $t );
2138                                 $closematch = preg_match(
2139                                         '/(<\\/table|<\\/blockquote|<\\/h1|<\\/h2|<\\/h3|<\\/h4|<\\/h5|<\\/h6|'.
2140                                         '<td|<th|<\\/?div|<hr|<\\/pre|<\\/p|'.$this->mUniqPrefix.'-pre|<\\/li|<\\/ul|<\\/ol|<\\/?center)/iS', $t );
2141                                 if ( $openmatch or $closematch ) {
2142                                         $paragraphStack = false;
2143                                         # TODO bug 5718: paragraph closed
2144                                         $output .= $this->closeParagraph();
2145                                         if ( $preOpenMatch and !$preCloseMatch ) {
2146                                                 $this->mInPre = true;
2147                                         }
2148                                         if ( $closematch ) {
2149                                                 $inBlockElem = false;
2150                                         } else {
2151                                                 $inBlockElem = true;
2152                                         }
2153                                 } else if ( !$inBlockElem && !$this->mInPre ) {
2154                                         if ( ' ' == $t{0} and ( $this->mLastSection == 'pre' or trim($t) != '' ) ) {
2155                                                 // pre
2156                                                 if ($this->mLastSection != 'pre') {
2157                                                         $paragraphStack = false;
2158                                                         $output .= $this->closeParagraph().'<pre>';
2159                                                         $this->mLastSection = 'pre';
2160                                                 }
2161                                                 $t = substr( $t, 1 );
2162                                         } else {
2163                                                 // paragraph
2164                                                 if ( '' == trim($t) ) {
2165                                                         if ( $paragraphStack ) {
2166                                                                 $output .= $paragraphStack.'<br />';
2167                                                                 $paragraphStack = false;
2168                                                                 $this->mLastSection = 'p';
2169                                                         } else {
2170                                                                 if ($this->mLastSection != 'p' ) {
2171                                                                         $output .= $this->closeParagraph();
2172                                                                         $this->mLastSection = '';
2173                                                                         $paragraphStack = '<p>';
2174                                                                 } else {
2175                                                                         $paragraphStack = '</p><p>';
2176                                                                 }
2177                                                         }
2178                                                 } else {
2179                                                         if ( $paragraphStack ) {
2180                                                                 $output .= $paragraphStack;
2181                                                                 $paragraphStack = false;
2182                                                                 $this->mLastSection = 'p';
2183                                                         } else if ($this->mLastSection != 'p') {
2184                                                                 $output .= $this->closeParagraph().'<p>';
2185                                                                 $this->mLastSection = 'p';
2186                                                         }
2187                                                 }
2188                                         }
2189                                 }
2190                                 wfProfileOut( "$fname-paragraph" );
2191                         }
2192                         // somewhere above we forget to get out of pre block (bug 785)
2193                         if($preCloseMatch && $this->mInPre) {
2194                                 $this->mInPre = false;
2195                         }
2196                         if ($paragraphStack === false) {
2197                                 $output .= $t."\n";
2198                         }
2199                 }
2200                 while ( $prefixLength ) {
2201                         $output .= $this->closeList( $pref2{$prefixLength-1} );
2202                         --$prefixLength;
2203                 }
2204                 if ( '' != $this->mLastSection ) {
2205                         $output .= '</' . $this->mLastSection . '>';
2206                         $this->mLastSection = '';
2207                 }
2208
2209                 wfProfileOut( $fname );
2210                 return $output;
2211         }
2212
2213         /**
2214          * Split up a string on ':', ignoring any occurences inside tags
2215          * to prevent illegal overlapping.
2216          * @param string $str the string to split
2217          * @param string &$before set to everything before the ':'
2218          * @param string &$after set to everything after the ':'
2219          * return string the position of the ':', or false if none found
2220          */
2221         function findColonNoLinks($str, &$before, &$after) {
2222                 $fname = 'Parser::findColonNoLinks';
2223                 wfProfileIn( $fname );
2224
2225                 $pos = strpos( $str, ':' );
2226                 if( $pos === false ) {
2227                         // Nothing to find!
2228                         wfProfileOut( $fname );
2229                         return false;
2230                 }
2231
2232                 $lt = strpos( $str, '<' );
2233                 if( $lt === false || $lt > $pos ) {
2234                         // Easy; no tag nesting to worry about
2235                         $before = substr( $str, 0, $pos );
2236                         $after = substr( $str, $pos+1 );
2237                         wfProfileOut( $fname );
2238                         return $pos;
2239                 }
2240
2241                 // Ugly state machine to walk through avoiding tags.
2242                 $state = MW_COLON_STATE_TEXT;
2243                 $stack = 0;
2244                 $len = strlen( $str );
2245                 for( $i = 0; $i < $len; $i++ ) {
2246                         $c = $str{$i};
2247
2248                         switch( $state ) {
2249                         // (Using the number is a performance hack for common cases)
2250                         case 0: // MW_COLON_STATE_TEXT:
2251                                 switch( $c ) {
2252                                 case "<":
2253                                         // Could be either a <start> tag or an </end> tag
2254                                         $state = MW_COLON_STATE_TAGSTART;
2255                                         break;
2256                                 case ":":
2257                                         if( $stack == 0 ) {
2258                                                 // We found it!
2259                                                 $before = substr( $str, 0, $i );
2260                                                 $after = substr( $str, $i + 1 );
2261                                                 wfProfileOut( $fname );
2262                                                 return $i;
2263                                         }
2264                                         // Embedded in a tag; don't break it.
2265                                         break;
2266                                 default:
2267                                         // Skip ahead looking for something interesting
2268                                         $colon = strpos( $str, ':', $i );
2269                                         if( $colon === false ) {
2270                                                 // Nothing else interesting
2271                                                 wfProfileOut( $fname );
2272                                                 return false;
2273                                         }
2274                                         $lt = strpos( $str, '<', $i );
2275                                         if( $stack === 0 ) {
2276                                                 if( $lt === false || $colon < $lt ) {
2277                                                         // We found it!
2278                                                         $before = substr( $str, 0, $colon );
2279                                                         $after = substr( $str, $colon + 1 );
2280                                                         wfProfileOut( $fname );
2281                                                         return $i;
2282                                                 }
2283                                         }
2284                                         if( $lt === false ) {
2285                                                 // Nothing else interesting to find; abort!
2286                                                 // We're nested, but there's no close tags left. Abort!
2287                                                 break 2;
2288                                         }
2289                                         // Skip ahead to next tag start
2290                                         $i = $lt;
2291                                         $state = MW_COLON_STATE_TAGSTART;
2292                                 }
2293                                 break;
2294                         case 1: // MW_COLON_STATE_TAG:
2295                                 // In a <tag>
2296                                 switch( $c ) {
2297                                 case ">":
2298                                         $stack++;
2299                                         $state = MW_COLON_STATE_TEXT;
2300                                         break;
2301                                 case "/":
2302                                         // Slash may be followed by >?
2303                                         $state = MW_COLON_STATE_TAGSLASH;
2304                                         break;
2305                                 default:
2306                                         // ignore
2307                                 }
2308                                 break;
2309                         case 2: // MW_COLON_STATE_TAGSTART:
2310                                 switch( $c ) {
2311                                 case "/":
2312                                         $state = MW_COLON_STATE_CLOSETAG;
2313                                         break;
2314                                 case "!":
2315                                         $state = MW_COLON_STATE_COMMENT;
2316                                         break;
2317                                 case ">":
2318                                         // Illegal early close? This shouldn't happen D:
2319                                         $state = MW_COLON_STATE_TEXT;
2320                                         break;
2321                                 default:
2322                                         $state = MW_COLON_STATE_TAG;
2323                                 }
2324                                 break;
2325                         case 3: // MW_COLON_STATE_CLOSETAG:
2326                                 // In a </tag>
2327                                 if( $c == ">" ) {
2328                                         $stack--;
2329                                         if( $stack < 0 ) {
2330                                                 wfDebug( "Invalid input in $fname; too many close tags\n" );
2331                                                 wfProfileOut( $fname );
2332                                                 return false;
2333                                         }
2334                                         $state = MW_COLON_STATE_TEXT;
2335                                 }
2336                                 break;
2337                         case MW_COLON_STATE_TAGSLASH:
2338                                 if( $c == ">" ) {
2339                                         // Yes, a self-closed tag <blah/>
2340                                         $state = MW_COLON_STATE_TEXT;
2341                                 } else {
2342                                         // Probably we're jumping the gun, and this is an attribute
2343                                         $state = MW_COLON_STATE_TAG;
2344                                 }
2345                                 break;
2346                         case 5: // MW_COLON_STATE_COMMENT:
2347                                 if( $c == "-" ) {
2348                                         $state = MW_COLON_STATE_COMMENTDASH;
2349                                 }
2350                                 break;
2351                         case MW_COLON_STATE_COMMENTDASH:
2352                                 if( $c == "-" ) {
2353                                         $state = MW_COLON_STATE_COMMENTDASHDASH;
2354                                 } else {
2355                                         $state = MW_COLON_STATE_COMMENT;
2356                                 }
2357                                 break;
2358                         case MW_COLON_STATE_COMMENTDASHDASH:
2359                                 if( $c == ">" ) {
2360                                         $state = MW_COLON_STATE_TEXT;
2361                                 } else {
2362                                         $state = MW_COLON_STATE_COMMENT;
2363                                 }
2364                                 break;
2365                         default:
2366                                 throw new MWException( "State machine error in $fname" );
2367                         }
2368                 }
2369                 if( $stack > 0 ) {
2370                         wfDebug( "Invalid input in $fname; not enough close tags (stack $stack, state $state)\n" );
2371                         return false;
2372                 }
2373                 wfProfileOut( $fname );
2374                 return false;
2375         }
2376
2377         /**
2378          * Return value of a magic variable (like PAGENAME)
2379          *
2380          * @private
2381          */
2382         function getVariableValue( $index ) {
2383                 global $wgContLang, $wgSitename, $wgServer, $wgServerName, $wgScriptPath;
2384
2385                 /**
2386                  * Some of these require message or data lookups and can be
2387                  * expensive to check many times.
2388                  */
2389                 static $varCache = array();
2390                 if ( wfRunHooks( 'ParserGetVariableValueVarCache', array( &$this, &$varCache ) ) ) {
2391                         if ( isset( $varCache[$index] ) ) {
2392                                 return $varCache[$index];
2393                         }
2394                 }
2395
2396                 $ts = time();
2397                 wfRunHooks( 'ParserGetVariableValueTs', array( &$this, &$ts ) );
2398
2399                 # Use the time zone
2400                 global $wgLocaltimezone;
2401                 if ( isset( $wgLocaltimezone ) ) {
2402                         $oldtz = getenv( 'TZ' );
2403                         putenv( 'TZ='.$wgLocaltimezone );
2404                 }
2405                 $localTimestamp = date( 'YmdHis', $ts );
2406                 $localMonth = date( 'm', $ts );
2407                 $localMonthName = date( 'n', $ts );
2408                 $localDay = date( 'j', $ts );
2409                 $localDay2 = date( 'd', $ts );
2410                 $localDayOfWeek = date( 'w', $ts );
2411                 $localWeek = date( 'W', $ts );
2412                 $localYear = date( 'Y', $ts );
2413                 $localHour = date( 'H', $ts );
2414                 if ( isset( $wgLocaltimezone ) ) {
2415                         putenv( 'TZ='.$oldtz );
2416                 }
2417
2418                 switch ( $index ) {
2419                         case 'currentmonth':
2420                                 return $varCache[$index] = $wgContLang->formatNum( date( 'm', $ts ) );
2421                         case 'currentmonthname':
2422                                 return $varCache[$index] = $wgContLang->getMonthName( date( 'n', $ts ) );
2423                         case 'currentmonthnamegen':
2424                                 return $varCache[$index] = $wgContLang->getMonthNameGen( date( 'n', $ts ) );
2425                         case 'currentmonthabbrev':
2426                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( date( 'n', $ts ) );
2427                         case 'currentday':
2428                                 return $varCache[$index] = $wgContLang->formatNum( date( 'j', $ts ) );
2429                         case 'currentday2':
2430                                 return $varCache[$index] = $wgContLang->formatNum( date( 'd', $ts ) );
2431                         case 'localmonth':
2432                                 return $varCache[$index] = $wgContLang->formatNum( $localMonth );
2433                         case 'localmonthname':
2434                                 return $varCache[$index] = $wgContLang->getMonthName( $localMonthName );
2435                         case 'localmonthnamegen':
2436                                 return $varCache[$index] = $wgContLang->getMonthNameGen( $localMonthName );
2437                         case 'localmonthabbrev':
2438                                 return $varCache[$index] = $wgContLang->getMonthAbbreviation( $localMonthName );
2439                         case 'localday':
2440                                 return $varCache[$index] = $wgContLang->formatNum( $localDay );
2441                         case 'localday2':
2442                                 return $varCache[$index] = $wgContLang->formatNum( $localDay2 );
2443                         case 'pagename':
2444                                 return $this->mTitle->getText();
2445                         case 'pagenamee':
2446                                 return $this->mTitle->getPartialURL();
2447                         case 'fullpagename':
2448                                 return $this->mTitle->getPrefixedText();
2449                         case 'fullpagenamee':
2450                                 return $this->mTitle->getPrefixedURL();
2451                         case 'subpagename':
2452                                 return $this->mTitle->getSubpageText();
2453                         case 'subpagenamee':
2454                                 return $this->mTitle->getSubpageUrlForm();
2455                         case 'basepagename':
2456                                 return $this->mTitle->getBaseText();
2457                         case 'basepagenamee':
2458                                 return wfUrlEncode( str_replace( ' ', '_', $this->mTitle->getBaseText() ) );
2459                         case 'talkpagename':
2460                                 if( $this->mTitle->canTalk() ) {
2461                                         $talkPage = $this->mTitle->getTalkPage();
2462                                         return $talkPage->getPrefixedText();
2463                                 } else {
2464                                         return '';
2465                                 }
2466                         case 'talkpagenamee':
2467                                 if( $this->mTitle->canTalk() ) {
2468                                         $talkPage = $this->mTitle->getTalkPage();
2469                                         return $talkPage->getPrefixedUrl();
2470                                 } else {
2471                                         return '';
2472                                 }
2473                         case 'subjectpagename':
2474                                 $subjPage = $this->mTitle->getSubjectPage();
2475                                 return $subjPage->getPrefixedText();
2476                         case 'subjectpagenamee':
2477                                 $subjPage = $this->mTitle->getSubjectPage();
2478                                 return $subjPage->getPrefixedUrl();
2479                         case 'revisionid':
2480                                 return $this->mRevisionId;
2481                         case 'revisionday':
2482                                 return intval( substr( $this->getRevisionTimestamp(), 6, 2 ) );
2483                         case 'revisionday2':
2484                                 return substr( $this->getRevisionTimestamp(), 6, 2 );
2485                         case 'revisionmonth':
2486                                 return intval( substr( $this->getRevisionTimestamp(), 4, 2 ) );
2487                         case 'revisionyear':
2488                                 return substr( $this->getRevisionTimestamp(), 0, 4 );
2489                         case 'revisiontimestamp':
2490                                 return $this->getRevisionTimestamp();
2491                         case 'namespace':
2492                                 return str_replace('_',' ',$wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2493                         case 'namespacee':
2494                                 return wfUrlencode( $wgContLang->getNsText( $this->mTitle->getNamespace() ) );
2495                         case 'talkspace':
2496                                 return $this->mTitle->canTalk() ? str_replace('_',' ',$this->mTitle->getTalkNsText()) : '';
2497                         case 'talkspacee':
2498                                 return $this->mTitle->canTalk() ? wfUrlencode( $this->mTitle->getTalkNsText() ) : '';
2499                         case 'subjectspace':
2500                                 return $this->mTitle->getSubjectNsText();
2501                         case 'subjectspacee':
2502                                 return( wfUrlencode( $this->mTitle->getSubjectNsText() ) );
2503                         case 'currentdayname':
2504                                 return $varCache[$index] = $wgContLang->getWeekdayName( date( 'w', $ts ) + 1 );
2505                         case 'currentyear':
2506                                 return $varCache[$index] = $wgContLang->formatNum( date( 'Y', $ts ), true );
2507                         case 'currenttime':
2508                                 return $varCache[$index] = $wgContLang->time( wfTimestamp( TS_MW, $ts ), false, false );
2509                         case 'currenthour':
2510                                 return $varCache[$index] = $wgContLang->formatNum( date( 'H', $ts ), true );
2511                         case 'currentweek':
2512                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2513                                 // int to remove the padding
2514                                 return $varCache[$index] = $wgContLang->formatNum( (int)date( 'W', $ts ) );
2515                         case 'currentdow':
2516                                 return $varCache[$index] = $wgContLang->formatNum( date( 'w', $ts ) );
2517                         case 'localdayname':
2518                                 return $varCache[$index] = $wgContLang->getWeekdayName( $localDayOfWeek + 1 );
2519                         case 'localyear':
2520                                 return $varCache[$index] = $wgContLang->formatNum( $localYear, true );
2521                         case 'localtime':
2522                                 return $varCache[$index] = $wgContLang->time( $localTimestamp, false, false );
2523                         case 'localhour':
2524                                 return $varCache[$index] = $wgContLang->formatNum( $localHour, true );
2525                         case 'localweek':
2526                                 // @bug 4594 PHP5 has it zero padded, PHP4 does not, cast to
2527                                 // int to remove the padding
2528                                 return $varCache[$index] = $wgContLang->formatNum( (int)$localWeek );
2529                         case 'localdow':
2530                                 return $varCache[$index] = $wgContLang->formatNum( $localDayOfWeek );
2531                         case 'numberofarticles':
2532                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::articles() );
2533                         case 'numberoffiles':
2534                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::images() );
2535                         case 'numberofusers':
2536                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::users() );
2537                         case 'numberofpages':
2538                                 return $varCache[$index] = $wgContLang->formatNum( SiteStats::pages() );
2539                         case 'numberofadmins':
2540                                 return $varCache[$index]  = $wgContLang->formatNum( SiteStats::admins() );
2541                         case 'currenttimestamp':
2542                                 return $varCache[$index] = wfTimestampNow();
2543                         case 'localtimestamp':
2544                                 return $varCache[$index] = $localTimestamp;
2545                         case 'currentversion':
2546                                 return $varCache[$index] = SpecialVersion::getVersion();
2547                         case 'sitename':
2548                                 return $wgSitename;
2549                         case 'server':
2550                                 return $wgServer;
2551                         case 'servername':
2552                                 return $wgServerName;
2553                         case 'scriptpath':
2554                                 return $wgScriptPath;
2555                         case 'directionmark':
2556                                 return $wgContLang->getDirMark();
2557                         case 'contentlanguage':
2558                                 global $wgContLanguageCode;
2559                                 return $wgContLanguageCode;
2560                         default:
2561                                 $ret = null;
2562                                 if ( wfRunHooks( 'ParserGetVariableValueSwitch', array( &$this, &$varCache, &$index, &$ret ) ) )
2563                                         return $ret;
2564                                 else
2565                                         return null;
2566                 }
2567         }
2568
2569         /**
2570          * initialise the magic variables (like CURRENTMONTHNAME)
2571          *
2572          * @private
2573          */
2574         function initialiseVariables() {
2575                 $fname = 'Parser::initialiseVariables';
2576                 wfProfileIn( $fname );
2577                 $variableIDs = MagicWord::getVariableIDs();
2578
2579                 $this->mVariables = array();
2580                 foreach ( $variableIDs as $id ) {
2581                         $mw =& MagicWord::get( $id );
2582                         $mw->addToArray( $this->mVariables, $id );
2583                 }
2584                 wfProfileOut( $fname );
2585         }
2586
2587         /**
2588          * parse any parentheses in format ((title|part|part))
2589          * and call callbacks to get a replacement text for any found piece
2590          *
2591          * @param string $text The text to parse
2592          * @param array $callbacks rules in form:
2593          *     '{' => array(                            # opening parentheses
2594          *                                      'end' => '}',   # closing parentheses
2595          *                                      'cb' => array(2 => callback,    # replacement callback to call if {{..}} is found
2596          *                                                                3 => callback         # replacement callback to call if {{{..}}} is found
2597          *                                                                )
2598          *                                      )
2599          *                                      'min' => 2,     # Minimum parenthesis count in cb
2600          *                                      'max' => 3,     # Maximum parenthesis count in cb
2601          * @private
2602          */
2603         function replace_callback ($text, $callbacks) {
2604                 wfProfileIn( __METHOD__ );
2605                 $openingBraceStack = array();   # this array will hold a stack of parentheses which are not closed yet
2606                 $lastOpeningBrace = -1;                 # last not closed parentheses
2607
2608                 $validOpeningBraces = implode( '', array_keys( $callbacks ) );
2609
2610                 $i = 0;
2611                 while ( $i < strlen( $text ) ) {
2612                         # Find next opening brace, closing brace or pipe
2613                         if ( $lastOpeningBrace == -1 ) {
2614                                 $currentClosing = '';
2615                                 $search = $validOpeningBraces;
2616                         } else {
2617                                 $currentClosing = $openingBraceStack[$lastOpeningBrace]['braceEnd'];
2618                                 $search = $validOpeningBraces . '|' . $currentClosing;
2619                         }
2620                         $rule = null;
2621                         $i += strcspn( $text, $search, $i );
2622                         if ( $i < strlen( $text ) ) {
2623                                 if ( $text[$i] == '|' ) {
2624                                         $found = 'pipe';
2625                                 } elseif ( $text[$i] == $currentClosing ) {
2626                                         $found = 'close';
2627                                 } elseif ( isset( $callbacks[$text[$i]] ) ) {
2628                                         $found = 'open';
2629                                         $rule = $callbacks[$text[$i]];
2630                                 } else {
2631                                         # Some versions of PHP have a strcspn which stops on null characters
2632                                         # Ignore and continue
2633                                         ++$i;
2634                                         continue;
2635                                 }
2636                         } else {
2637                                 # All done
2638                                 break;
2639                         }
2640
2641                         if ( $found == 'open' ) {
2642                                 # found opening brace, let's add it to parentheses stack
2643                                 $piece = array('brace' => $text[$i],
2644                                                            'braceEnd' => $rule['end'],
2645                                                            'title' => '',
2646                                                            'parts' => null);
2647
2648                                 # count opening brace characters
2649                                 $piece['count'] = strspn( $text, $piece['brace'], $i );
2650                                 $piece['startAt'] = $piece['partStart'] = $i + $piece['count'];
2651                                 $i += $piece['count'];
2652
2653                                 # we need to add to stack only if opening brace count is enough for one of the rules
2654                                 if ( $piece['count'] >= $rule['min'] ) {
2655                                         $lastOpeningBrace ++;
2656                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2657                                 }
2658                         } elseif ( $found == 'close' ) {
2659                                 # lets check if it is enough characters for closing brace
2660                                 $maxCount = $openingBraceStack[$lastOpeningBrace]['count'];
2661                                 $count = strspn( $text, $text[$i], $i, $maxCount );
2662
2663                                 # check for maximum matching characters (if there are 5 closing
2664                                 # characters, we will probably need only 3 - depending on the rules)
2665                                 $matchingCount = 0;
2666                                 $matchingCallback = null;
2667                                 $cbType = $callbacks[$openingBraceStack[$lastOpeningBrace]['brace']];
2668                                 if ( $count > $cbType['max'] ) {
2669                                         # The specified maximum exists in the callback array, unless the caller
2670                                         # has made an error
2671                                         $matchingCount = $cbType['max'];
2672                                 } else {
2673                                         # Count is less than the maximum
2674                                         # Skip any gaps in the callback array to find the true largest match
2675                                         # Need to use array_key_exists not isset because the callback can be null
2676                                         $matchingCount = $count;
2677                                         while ( $matchingCount > 0 && !array_key_exists( $matchingCount, $cbType['cb'] ) ) {
2678                                                 --$matchingCount;
2679                                         }
2680                                 }
2681
2682                                 if ($matchingCount <= 0) {
2683                                         $i += $count;
2684                                         continue;
2685                                 }
2686                                 $matchingCallback = $cbType['cb'][$matchingCount];
2687
2688                                 # let's set a title or last part (if '|' was found)
2689                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2690                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2691                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2692                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2693                                 } else {
2694                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2695                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2696                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2697                                 }
2698
2699                                 $pieceStart = $openingBraceStack[$lastOpeningBrace]['startAt'] - $matchingCount;
2700                                 $pieceEnd = $i + $matchingCount;
2701
2702                                 if( is_callable( $matchingCallback ) ) {
2703                                         $cbArgs = array (
2704                                                                          'text' => substr($text, $pieceStart, $pieceEnd - $pieceStart),
2705                                                                          'title' => trim($openingBraceStack[$lastOpeningBrace]['title']),
2706                                                                          'parts' => $openingBraceStack[$lastOpeningBrace]['parts'],
2707                                                                          'lineStart' => (($pieceStart > 0) && ($text[$pieceStart-1] == "\n")),
2708                                                                          );
2709                                         # finally we can call a user callback and replace piece of text
2710                                         $replaceWith = call_user_func( $matchingCallback, $cbArgs );
2711                                         $text = substr($text, 0, $pieceStart) . $replaceWith . substr($text, $pieceEnd);
2712                                         $i = $pieceStart + strlen($replaceWith);
2713                                 } else {
2714                                         # null value for callback means that parentheses should be parsed, but not replaced
2715                                         $i += $matchingCount;
2716                                 }
2717
2718                                 # reset last opening parentheses, but keep it in case there are unused characters
2719                                 $piece = array('brace' => $openingBraceStack[$lastOpeningBrace]['brace'],
2720                                                            'braceEnd' => $openingBraceStack[$lastOpeningBrace]['braceEnd'],
2721                                                            'count' => $openingBraceStack[$lastOpeningBrace]['count'],
2722                                                            'title' => '',
2723                                                            'parts' => null,
2724                                                            'startAt' => $openingBraceStack[$lastOpeningBrace]['startAt']);
2725                                 $openingBraceStack[$lastOpeningBrace--] = null;
2726
2727                                 if ($matchingCount < $piece['count']) {
2728                                         $piece['count'] -= $matchingCount;
2729                                         $piece['startAt'] -= $matchingCount;
2730                                         $piece['partStart'] = $piece['startAt'];
2731                                         # do we still qualify for any callback with remaining count?
2732                                         $currentCbList = $callbacks[$piece['brace']]['cb'];
2733                                         while ( $piece['count'] ) {
2734                                                 if ( array_key_exists( $piece['count'], $currentCbList ) ) {
2735                                                         $lastOpeningBrace++;
2736                                                         $openingBraceStack[$lastOpeningBrace] = $piece;
2737                                                         break;
2738                                                 }
2739                                                 --$piece['count'];
2740                                         }
2741                                 }
2742                         } elseif ( $found == 'pipe' ) {
2743                                 # lets set a title if it is a first separator, or next part otherwise
2744                                 if (null === $openingBraceStack[$lastOpeningBrace]['parts']) {
2745                                         $openingBraceStack[$lastOpeningBrace]['title'] =
2746                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2747                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2748                                         $openingBraceStack[$lastOpeningBrace]['parts'] = array();
2749                                 } else {
2750                                         $openingBraceStack[$lastOpeningBrace]['parts'][] =
2751                                                 substr($text, $openingBraceStack[$lastOpeningBrace]['partStart'],
2752                                                 $i - $openingBraceStack[$lastOpeningBrace]['partStart']);
2753                                 }
2754                                 $openingBraceStack[$lastOpeningBrace]['partStart'] = ++$i;
2755                         }
2756                 }
2757
2758                 wfProfileOut( __METHOD__ );
2759                 return $text;
2760         }
2761
2762         /**
2763          * Replace magic variables, templates, and template arguments
2764          * with the appropriate text. Templates are substituted recursively,
2765          * taking care to avoid infinite loops.
2766          *
2767          * Note that the substitution depends on value of $mOutputType:
2768          *  OT_WIKI: only {{subst:}} templates
2769          *  OT_MSG: only magic variables
2770          *  OT_HTML: all templates and magic variables
2771          *
2772          * @param string $tex The text to transform
2773          * @param array $args Key-value pairs representing template parameters to substitute
2774          * @param bool $argsOnly Only do argument (triple-brace) expansion, not double-brace expansion
2775          * @private
2776          */
2777         function replaceVariables( $text, $args = array(), $argsOnly = false ) {
2778                 # Prevent too big inclusions
2779                 if( strlen( $text ) > $this->mOptions->getMaxIncludeSize() ) {
2780                         return $text;
2781                 }
2782
2783                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2784                 wfProfileIn( $fname );
2785
2786                 # This function is called recursively. To keep track of arguments we need a stack:
2787                 array_push( $this->mArgStack, $args );
2788
2789                 $braceCallbacks = array();
2790                 if ( !$argsOnly ) {
2791                         $braceCallbacks[2] = array( &$this, 'braceSubstitution' );
2792                 }
2793                 if ( $this->mOutputType != OT_MSG ) {
2794                         $braceCallbacks[3] = array( &$this, 'argSubstitution' );
2795                 }
2796                 if ( $braceCallbacks ) {
2797                         $callbacks = array(
2798                                 '{' => array(
2799                                         'end' => '}',
2800                                         'cb' => $braceCallbacks,
2801                                         'min' => $argsOnly ? 3 : 2,
2802                                         'max' => isset( $braceCallbacks[3] ) ? 3 : 2,
2803                                 ),
2804                                 '[' => array(
2805                                         'end' => ']',
2806                                         'cb' => array(2=>null),
2807                                         'min' => 2,
2808                                         'max' => 2,
2809                                 )
2810                         );
2811                         $text = $this->replace_callback ($text, $callbacks);
2812
2813                         array_pop( $this->mArgStack );
2814                 }
2815                 wfProfileOut( $fname );
2816                 return $text;
2817         }
2818
2819         /**
2820          * Replace magic variables
2821          * @private
2822          */
2823         function variableSubstitution( $matches ) {
2824                 global $wgContLang;
2825                 $fname = 'Parser::variableSubstitution';
2826                 $varname = $wgContLang->lc($matches[1]);
2827                 wfProfileIn( $fname );
2828                 $skip = false;
2829                 if ( $this->mOutputType == OT_WIKI ) {
2830                         # Do only magic variables prefixed by SUBST
2831                         $mwSubst =& MagicWord::get( 'subst' );
2832                         if (!$mwSubst->matchStartAndRemove( $varname ))
2833                                 $skip = true;
2834                         # Note that if we don't substitute the variable below,
2835                         # we don't remove the {{subst:}} magic word, in case
2836                         # it is a template rather than a magic variable.
2837                 }
2838                 if ( !$skip && array_key_exists( $varname, $this->mVariables ) ) {
2839                         $id = $this->mVariables[$varname];
2840                         # Now check if we did really match, case sensitive or not
2841                         $mw =& MagicWord::get( $id );
2842                         if ($mw->match($matches[1])) {
2843                                 $text = $this->getVariableValue( $id );
2844                                 $this->mOutput->mContainsOldMagic = true;
2845                         } else {
2846                                 $text = $matches[0];
2847                         }
2848                 } else {
2849                         $text = $matches[0];
2850                 }
2851                 wfProfileOut( $fname );
2852                 return $text;
2853         }
2854
2855
2856         /// Clean up argument array - refactored in 1.9 so parserfunctions can use it, too.
2857         static function createAssocArgs( $args ) {
2858                 $assocArgs = array();
2859                 $index = 1;
2860                 foreach( $args as $arg ) {
2861                         $eqpos = strpos( $arg, '=' );
2862                         if ( $eqpos === false ) {
2863                                 $assocArgs[$index++] = $arg;
2864                         } else {
2865                                 $name = trim( substr( $arg, 0, $eqpos ) );
2866                                 $value = trim( substr( $arg, $eqpos+1 ) );
2867                                 if ( $value === false ) {
2868                                         $value = '';
2869                                 }
2870                                 if ( $name !== false ) {
2871                                         $assocArgs[$name] = $value;
2872                                 }
2873                         }
2874                 }
2875
2876                 return $assocArgs;
2877         }
2878
2879         /**
2880          * Return the text of a template, after recursively
2881          * replacing any variables or templates within the template.
2882          *
2883          * @param array $piece The parts of the template
2884          *  $piece['text']: matched text
2885          *  $piece['title']: the title, i.e. the part before the |
2886          *  $piece['parts']: the parameter array
2887          * @return string the text of the template
2888          * @private
2889          */
2890         function braceSubstitution( $piece ) {
2891                 global $wgContLang, $wgLang, $wgAllowDisplayTitle, $wgNonincludableNamespaces;
2892                 $fname = __METHOD__ /*. '-L' . count( $this->mArgStack )*/;
2893                 wfProfileIn( $fname );
2894                 wfProfileIn( __METHOD__.'-setup' );
2895
2896                 # Flags
2897                 $found = false;             # $text has been filled
2898                 $nowiki = false;            # wiki markup in $text should be escaped
2899                 $noparse = false;           # Unsafe HTML tags should not be stripped, etc.
2900                 $noargs = false;            # Don't replace triple-brace arguments in $text
2901                 $replaceHeadings = false;   # Make the edit section links go to the template not the article
2902                 $headingOffset = 0;         # Skip headings when number, to account for those that weren't transcluded.
2903                 $isHTML = false;            # $text is HTML, armour it against wikitext transformation
2904                 $forceRawInterwiki = false; # Force interwiki transclusion to be done in raw mode not rendered
2905
2906                 # Title object, where $text came from
2907                 $title = NULL;
2908
2909                 $linestart = '';
2910
2911
2912                 # $part1 is the bit before the first |, and must contain only title characters
2913                 # $args is a list of arguments, starting from index 0, not including $part1
2914
2915                 $titleText = $part1 = $piece['title'];
2916                 # If the third subpattern matched anything, it will start with |
2917
2918                 if (null == $piece['parts']) {
2919                         $replaceWith = $this->variableSubstitution (array ($piece['text'], $piece['title']));
2920                         if ($replaceWith != $piece['text']) {
2921                                 $text = $replaceWith;
2922                                 $found = true;
2923                                 $noparse = true;
2924                                 $noargs = true;
2925                         }
2926                 }
2927
2928                 $args = (null == $piece['parts']) ? array() : $piece['parts'];
2929                 wfProfileOut( __METHOD__.'-setup' );
2930
2931                 # SUBST
2932                 wfProfileIn( __METHOD__.'-modifiers' );
2933                 if ( !$found ) {
2934                         $mwSubst =& MagicWord::get( 'subst' );
2935                         if ( $mwSubst->matchStartAndRemove( $part1 ) xor $this->ot['wiki'] ) {
2936                                 # One of two possibilities is true:
2937                                 # 1) Found SUBST but not in the PST phase
2938                                 # 2) Didn't find SUBST and in the PST phase
2939                                 # In either case, return without further processing
2940                                 $text = $piece['text'];
2941                                 $found = true;
2942                                 $noparse = true;
2943                                 $noargs = true;
2944                         }
2945                 }
2946
2947                 # MSG, MSGNW and RAW
2948                 if ( !$found ) {
2949                         # Check for MSGNW:
2950                         $mwMsgnw =& MagicWord::get( 'msgnw' );
2951                         if ( $mwMsgnw->matchStartAndRemove( $part1 ) ) {
2952                                 $nowiki = true;
2953                         } else {
2954                                 # Remove obsolete MSG:
2955                                 $mwMsg =& MagicWord::get( 'msg' );
2956                                 $mwMsg->matchStartAndRemove( $part1 );
2957                         }
2958
2959                         # Check for RAW:
2960                         $mwRaw =& MagicWord::get( 'raw' );
2961                         if ( $mwRaw->matchStartAndRemove( $part1 ) ) {
2962                                 $forceRawInterwiki = true;
2963                         }
2964                 }
2965                 wfProfileOut( __METHOD__.'-modifiers' );
2966
2967                 //save path level before recursing into functions & templates.
2968                 $lastPathLevel = $this->mTemplatePath;
2969
2970                 # Parser functions
2971                 if ( !$found ) {
2972                         wfProfileIn( __METHOD__ . '-pfunc' );
2973
2974                         $colonPos = strpos( $part1, ':' );
2975                         if ( $colonPos !== false ) {
2976                                 # Case sensitive functions
2977                                 $function = substr( $part1, 0, $colonPos );
2978                                 if ( isset( $this->mFunctionSynonyms[1][$function] ) ) {
2979                                         $function = $this->mFunctionSynonyms[1][$function];
2980                                 } else {
2981                                         # Case insensitive functions
2982                                         $function = strtolower( $function );
2983                                         if ( isset( $this->mFunctionSynonyms[0][$function] ) ) {
2984                                                 $function = $this->mFunctionSynonyms[0][$function];
2985                                         } else {
2986                                                 $function = false;
2987                                         }
2988                                 }
2989                                 if ( $function ) {
2990                                         $funcArgs = array_map( 'trim', $args );
2991                                         $funcArgs = array_merge( array( &$this, trim( substr( $part1, $colonPos + 1 ) ) ), $funcArgs );
2992                                         $result = call_user_func_array( $this->mFunctionHooks[$function], $funcArgs );
2993                                         $found = true;
2994
2995                                         // The text is usually already parsed, doesn't need triple-brace tags expanded, etc.
2996                                         //$noargs = true;
2997                                         //$noparse = true;
2998
2999                                         if ( is_array( $result ) ) {
3000                                                 if ( isset( $result[0] ) ) {
3001                                                         $text = $linestart . $result[0];
3002                                                         unset( $result[0] );
3003                                                 }
3004
3005                                                 // Extract flags into the local scope
3006                                                 // This allows callers to set flags such as nowiki, noparse, found, etc.
3007                                                 extract( $result );
3008                                         } else {
3009                                                 $text = $linestart . $result;
3010                                         }
3011                                 }
3012                         }
3013                         wfProfileOut( __METHOD__ . '-pfunc' );
3014                 }
3015
3016                 # Template table test
3017
3018                 # Did we encounter this template already? If yes, it is in the cache
3019                 # and we need to check for loops.
3020                 if ( !$found && isset( $this->mTemplates[$piece['title']] ) ) {
3021                         $found = true;
3022
3023                         # Infinite loop test
3024                         if ( isset( $this->mTemplatePath[$part1] ) ) {
3025                                 $noparse = true;
3026                                 $noargs = true;
3027                                 $found = true;
3028                                 $text = $linestart .
3029                                         "[[$part1]]<!-- WARNING: template loop detected -->";
3030                                 wfDebug( __METHOD__.": template loop broken at '$part1'\n" );
3031                         } else {
3032                                 # set $text to cached message.
3033                                 $text = $linestart . $this->mTemplates[$piece['title']];
3034                         }
3035                 }
3036
3037                 # Load from database
3038                 if ( !$found ) {
3039                         wfProfileIn( __METHOD__ . '-loadtpl' );
3040                         $ns = NS_TEMPLATE;
3041                         # declaring $subpage directly in the function call
3042                         # does not work correctly with references and breaks
3043                         # {{/subpage}}-style inclusions
3044                         $subpage = '';
3045                         $part1 = $this->maybeDoSubpageLink( $part1, $subpage );
3046                         if ($subpage !== '') {
3047                                 $ns = $this->mTitle->getNamespace();
3048                         }
3049                         $title = Title::newFromText( $part1, $ns );
3050
3051
3052                         if ( !is_null( $title ) ) {
3053                                 $titleText = $title->getPrefixedText();
3054                                 # Check for language variants if the template is not found
3055                                 if($wgContLang->hasVariants() && $title->getArticleID() == 0){
3056                                         $wgContLang->findVariantLink($part1, $title);
3057                                 }
3058
3059                                 if ( !$title->isExternal() ) {
3060                                         if ( $title->getNamespace() == NS_SPECIAL && $this->mOptions->getAllowSpecialInclusion() && $this->ot['html'] ) {
3061                                                 $text = SpecialPage::capturePath( $title );
3062                                                 if ( is_string( $text ) ) {
3063                                                         $found = true;
3064                                                         $noparse = true;
3065                                                         $noargs = true;
3066                                                         $isHTML = true;
3067                                                         $this->disableCache();
3068                                                 }
3069                                         } else if ( $wgNonincludableNamespaces && in_array( $title->getNamespace(), $wgNonincludableNamespaces ) ) {
3070                                                 $found = false; //access denied
3071                                                 wfDebug( "$fname: template inclusion denied for " . $title->getPrefixedDBkey() );
3072                                         } else {
3073                                                 $articleContent = $this->fetchTemplate( $title );
3074                                                 if ( $articleContent !== false ) {
3075                                                         $found = true;
3076                                                         $text = $articleContent;
3077                                                         $replaceHeadings = true;
3078                                                 }
3079                                         }
3080
3081                                         # If the title is valid but undisplayable, make a link to it
3082                                         if ( !$found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3083                                                 $text = "[[:$titleText]]";
3084                                                 $found = true;
3085                                         }
3086                                 } elseif ( $title->isTrans() ) {
3087                                         // Interwiki transclusion
3088                                         if ( $this->ot['html'] && !$forceRawInterwiki ) {
3089                                                 $text = $this->interwikiTransclude( $title, 'render' );
3090                                                 $isHTML = true;
3091                                                 $noparse = true;
3092                                         } else {
3093                                                 $text = $this->interwikiTransclude( $title, 'raw' );
3094                                                 $replaceHeadings = true;
3095                                         }
3096                                         $found = true;
3097                                 }
3098
3099                                 # Template cache array insertion
3100                                 # Use the original $piece['title'] not the mangled $part1, so that
3101                                 # modifiers such as RAW: produce separate cache entries
3102                                 if( $found ) {
3103                                         if( $isHTML ) {
3104                                                 // A special page; don't store it in the template cache.
3105                                         } else {
3106                                                 $this->mTemplates[$piece['title']] = $text;
3107                                         }
3108                                         $text = $linestart . $text;
3109                                 }
3110                         }
3111                         wfProfileOut( __METHOD__ . '-loadtpl' );
3112                 }
3113
3114                 if ( $found && !$this->incrementIncludeSize( 'pre-expand', strlen( $text ) ) ) {
3115                         # Error, oversize inclusion
3116                         $text = $linestart .
3117                                 "[[$titleText]]<!-- WARNING: template omitted, pre-expand include size too large -->";
3118                         $noparse = true;
3119                         $noargs = true;
3120                 }
3121
3122                 # Recursive parsing, escaping and link table handling
3123                 # Only for HTML output
3124                 if ( $nowiki && $found && ( $this->ot['html'] || $this->ot['pre'] ) ) {
3125                         $text = wfEscapeWikiText( $text );
3126                 } elseif ( !$this->ot['msg'] && $found ) {
3127                         if ( $noargs ) {
3128                                 $assocArgs = array();
3129                         } else {
3130                                 # Clean up argument array
3131                                 $assocArgs = self::createAssocArgs($args);
3132                                 # Add a new element to the templace recursion path
3133                                 $this->mTemplatePath[$part1] = 1;
3134                         }
3135
3136                         if ( !$noparse ) {
3137                                 # If there are any <onlyinclude> tags, only include them
3138                                 if ( in_string( '<onlyinclude>', $text ) && in_string( '</onlyinclude>', $text ) ) {
3139                                         $replacer = new OnlyIncludeReplacer;
3140                                         StringUtils::delimiterReplaceCallback( '<onlyinclude>', '</onlyinclude>',
3141                                                 array( &$replacer, 'replace' ), $text );
3142                                         $text = $replacer->output;
3143                                 }
3144                                 # Remove <noinclude> sections and <includeonly> tags
3145                                 $text = StringUtils::delimiterReplace( '<noinclude>', '</noinclude>', '', $text );
3146                                 $text = strtr( $text, array( '<includeonly>' => '' , '</includeonly>' => '' ) );
3147
3148                                 if( $this->ot['html'] || $this->ot['pre'] ) {
3149                                         # Strip <nowiki>, <pre>, etc.
3150                                         $text = $this->strip( $text, $this->mStripState );
3151                                         if ( $this->ot['html'] ) {
3152                                                 $text = Sanitizer::removeHTMLtags( $text, array( &$this, 'replaceVariables' ), $assocArgs );
3153                                         } elseif ( $this->ot['pre'] && $this->mOptions->getRemoveComments() ) {
3154                                                 $text = Sanitizer::removeHTMLcomments( $text );
3155                                         }
3156                                 }
3157                                 $text = $this->replaceVariables( $text, $assocArgs );
3158
3159                                 # If the template begins with a table or block-level
3160                                 # element, it should be treated as beginning a new line.
3161                                 if (!$piece['lineStart'] && preg_match('/^({\\||:|;|#|\*)/', $text)) /*}*/{
3162                                         $text = "\n" . $text;
3163                                 }
3164                         } elseif ( !$noargs ) {
3165                                 # $noparse and !$noargs
3166                                 # Just replace the arguments, not any double-brace items
3167                                 # This is used for rendered interwiki transclusion
3168                                 $text = $this->replaceVariables( $text, $assocArgs, true );
3169                         }
3170                 }
3171                 # Prune lower levels off the recursion check path
3172                 $this->mTemplatePath = $lastPathLevel;
3173
3174                 if ( $found && !$this->incrementIncludeSize( 'post-expand', strlen( $text ) ) ) {
3175                         # Error, oversize inclusion
3176                         $text = $linestart .
3177                                 "[[$titleText]]<!-- WARNING: template omitted, post-expand include size too large -->";
3178                         $noparse = true;
3179                         $noargs = true;
3180                 }
3181
3182                 if ( !$found ) {
3183                         wfProfileOut( $fname );
3184                         return $piece['text'];
3185                 } else {
3186                         wfProfileIn( __METHOD__ . '-placeholders' );
3187                         if ( $isHTML ) {
3188                                 # Replace raw HTML by a placeholder
3189                                 # Add a blank line preceding, to prevent it from mucking up
3190                                 # immediately preceding headings
3191                                 $text = "\n\n" . $this->insertStripItem( $text, $this->mStripState );
3192                         } else {
3193                                 # replace ==section headers==
3194                                 # XXX this needs to go away once we have a better parser.
3195                                 if ( !$this->ot['wiki'] && !$this->ot['pre'] && $replaceHeadings ) {
3196                                         if( !is_null( $title ) )
3197                                                 $encodedname = base64_encode($title->getPrefixedDBkey());
3198                                         else
3199                                                 $encodedname = base64_encode("");
3200                                         $m = preg_split('/(^={1,6}.*?={1,6}\s*?$)/m', $text, -1,
3201                                                 PREG_SPLIT_DELIM_CAPTURE);
3202                                         $text = '';
3203                                         $nsec = $headingOffset;
3204                                         for( $i = 0; $i < count($m); $i += 2 ) {
3205                                                 $text .= $m[$i];
3206                                                 if (!isset($m[$i + 1]) || $m[$i + 1] == "") continue;
3207                                                 $hl = $m[$i + 1];
3208                                                 if( strstr($hl, "<!--MWTEMPLATESECTION") ) {
3209                                                         $text .= $hl;
3210                                                         continue;
3211                                                 }
3212                                                 $m2 = array();
3213                                                 preg_match('/^(={1,6})(.*?)(={1,6})\s*?$/m', $hl, $m2);
3214                                                 $text .= $m2[1] . $m2[2] . "<!--MWTEMPLATESECTION="
3215                                                         . $encodedname . "&" . base64_encode("$nsec") . "-->" . $m2[3];
3216
3217                                                 $nsec++;
3218                                         }
3219                                 }
3220                         }
3221                         wfProfileOut( __METHOD__ . '-placeholders' );
3222                 }
3223
3224                 # Prune lower levels off the recursion check path
3225                 $this->mTemplatePath = $lastPathLevel;
3226
3227                 if ( !$found ) {
3228                         wfProfileOut( $fname );
3229                         return $piece['text'];
3230                 } else {
3231                         wfProfileOut( $fname );
3232                         return $text;
3233                 }
3234         }
3235
3236         /**
3237          * Fetch the unparsed text of a template and register a reference to it.
3238          */
3239         function fetchTemplate( $title ) {
3240                 $text = false;
3241                 // Loop to fetch the article, with up to 1 redirect
3242                 for ( $i = 0; $i < 2 && is_object( $title ); $i++ ) {
3243                         $rev = Revision::newFromTitle( $title );
3244                         $this->mOutput->addTemplate( $title, $title->getArticleID() );
3245                         if ( $rev ) {
3246                                 $text = $rev->getText();
3247                         } elseif( $title->getNamespace() == NS_MEDIAWIKI ) {
3248                                 global $wgLang;
3249                                 $message = $wgLang->lcfirst( $title->getText() );
3250                                 $text = wfMsgForContentNoTrans( $message );
3251                                 if( wfEmptyMsg( $message, $text ) ) {
3252                                         $text = false;
3253                                         break;
3254                                 }
3255                         } else {
3256                                 break;
3257                         }
3258                         if ( $text === false ) {
3259                                 break;
3260                         }
3261                         // Redirect?
3262                         $title = Title::newFromRedirect( $text );
3263                 }
3264                 return $text;
3265         }
3266
3267         /**
3268          * Transclude an interwiki link.
3269          */
3270         function interwikiTransclude( $title, $action ) {
3271                 global $wgEnableScaryTranscluding;
3272
3273                 if (!$wgEnableScaryTranscluding)
3274                         return wfMsg('scarytranscludedisabled');
3275
3276                 $url = $title->getFullUrl( "action=$action" );
3277
3278                 if (strlen($url) > 255)
3279                         return wfMsg('scarytranscludetoolong');
3280                 return $this->fetchScaryTemplateMaybeFromCache($url);
3281         }
3282
3283         function fetchScaryTemplateMaybeFromCache($url) {
3284                 global $wgTranscludeCacheExpiry;
3285                 $dbr = wfGetDB(DB_SLAVE);
3286                 $obj = $dbr->selectRow('transcache', array('tc_time', 'tc_contents'),
3287                                 array('tc_url' => $url));
3288                 if ($obj) {
3289                         $time = $obj->tc_time;
3290                         $text = $obj->tc_contents;
3291                         if ($time && time() < $time + $wgTranscludeCacheExpiry ) {
3292                                 return $text;
3293                         }
3294                 }
3295
3296                 $text = Http::get($url);
3297                 if (!$text)
3298                         return wfMsg('scarytranscludefailed', $url);
3299
3300                 $dbw = wfGetDB(DB_MASTER);
3301                 $dbw->replace('transcache', array('tc_url'), array(
3302                         'tc_url' => $url,
3303                         'tc_time' => time(),
3304                         'tc_contents' => $text));
3305                 return $text;
3306         }
3307
3308
3309         /**
3310          * Triple brace replacement -- used for template arguments
3311          * @private
3312          */
3313         function argSubstitution( $matches ) {
3314                 $arg = trim( $matches['title'] );
3315                 $text = $matches['text'];
3316                 $inputArgs = end( $this->mArgStack );
3317
3318                 if ( array_key_exists( $arg, $inputArgs ) ) {
3319                         $text = $inputArgs[$arg];
3320                 } else if (($this->mOutputType == OT_HTML || $this->mOutputType == OT_PREPROCESS ) &&
3321                 null != $matches['parts'] && count($matches['parts']) > 0) {
3322                         $text = $matches['parts'][0];
3323                 }
3324                 if ( !$this->incrementIncludeSize( 'arg', strlen( $text ) ) ) {
3325                         $text = $matches['text'] .
3326                                 '<!-- WARNING: argument omitted, expansion size too large -->';
3327                 }
3328
3329                 return $text;
3330         }
3331
3332         /**
3333          * Increment an include size counter
3334          *
3335          * @param string $type The type of expansion
3336          * @param integer $size The size of the text
3337          * @return boolean False if this inclusion would take it over the maximum, true otherwise
3338          */
3339         function incrementIncludeSize( $type, $size ) {
3340                 if ( $this->mIncludeSizes[$type] + $size > $this->mOptions->getMaxIncludeSize() ) {
3341                         return false;
3342                 } else {
3343                         $this->mIncludeSizes[$type] += $size;
3344                         return true;
3345                 }
3346         }
3347
3348         /**
3349          * Detect __NOGALLERY__ magic word and set a placeholder
3350          */
3351         function stripNoGallery( &$text ) {
3352                 # if the string __NOGALLERY__ (not case-sensitive) occurs in the HTML,
3353                 # do not add TOC
3354                 $mw = MagicWord::get( 'nogallery' );
3355                 $this->mOutput->mNoGallery = $mw->matchAndRemove( $text ) ;
3356         }
3357
3358         /**
3359          * Detect __TOC__ magic word and set a placeholder
3360          */
3361         function stripToc( $text ) {
3362                 # if the string __NOTOC__ (not case-sensitive) occurs in the HTML,
3363                 # do not add TOC
3364                 $mw = MagicWord::get( 'notoc' );
3365                 if( $mw->matchAndRemove( $text ) ) {
3366                         $this->mShowToc = false;
3367                 }
3368
3369                 $mw = MagicWord::get( 'toc' );
3370                 if( $mw->match( $text ) ) {
3371                         $this->mShowToc = true;
3372                         $this->mForceTocPosition = true;
3373
3374                         // Set a placeholder. At the end we'll fill it in with the TOC.
3375                         $text = $mw->replace( '<!--MWTOC-->', $text, 1 );
3376
3377                         // Only keep the first one.
3378                         $text = $mw->replace( '', $text );
3379                 }
3380                 return $text;
3381         }
3382
3383         /**
3384          * This function accomplishes several tasks:
3385          * 1) Auto-number headings if that option is enabled
3386          * 2) Add an [edit] link to sections for logged in users who have enabled the option
3387          * 3) Add a Table of contents on the top for users who have enabled the option
3388          * 4) Auto-anchor headings
3389          *
3390          * It loops through all headlines, collects the necessary data, then splits up the
3391          * string and re-inserts the newly formatted headlines.
3392          *
3393          * @param string $text
3394          * @param boolean $isMain
3395          * @private
3396          */
3397         function formatHeadings( $text, $isMain=true ) {
3398                 global $wgMaxTocLevel, $wgContLang;
3399
3400                 $doNumberHeadings = $this->mOptions->getNumberHeadings();
3401                 if( !$this->mTitle->quickUserCan( 'edit' ) ) {
3402                         $showEditLink = 0;
3403                 } else {
3404                         $showEditLink = $this->mOptions->getEditSection();
3405                 }
3406
3407                 # Inhibit editsection links if requested in the page
3408                 $esw =& MagicWord::get( 'noeditsection' );
3409                 if( $esw->matchAndRemove( $text ) ) {
3410                         $showEditLink = 0;
3411                 }
3412
3413                 # Get all headlines for numbering them and adding funky stuff like [edit]
3414                 # links - this is for later, but we need the number of headlines right now
3415                 $matches = array();
3416                 $numMatches = preg_match_all( '/<H(?P<level>[1-6])(?P<attrib>.*?'.'>)(?P<header>.*?)<\/H[1-6] *>/i', $text, $matches );
3417
3418                 # if there are fewer than 4 headlines in the article, do not show TOC
3419                 # unless it's been explicitly enabled.
3420                 $enoughToc = $this->mShowToc &&
3421                         (($numMatches >= 4) || $this->mForceTocPosition);
3422
3423                 # Allow user to stipulate that a page should have a "new section"
3424                 # link added via __NEWSECTIONLINK__
3425                 $mw =& MagicWord::get( 'newsectionlink' );
3426                 if( $mw->matchAndRemove( $text ) )
3427                         $this->mOutput->setNewSection( true );
3428
3429                 # if the string __FORCETOC__ (not case-sensitive) occurs in the HTML,
3430                 # override above conditions and always show TOC above first header
3431                 $mw =& MagicWord::get( 'forcetoc' );
3432                 if ($mw->matchAndRemove( $text ) ) {
3433                         $this->mShowToc = true;
3434                         $enoughToc = true;
3435                 }
3436
3437                 # Never ever show TOC if no headers
3438                 if( $numMatches < 1 ) {
3439                         $enoughToc = false;
3440                 }
3441
3442                 # We need this to perform operations on the HTML
3443                 $sk = $this->mOptions->getSkin();
3444
3445                 # headline counter
3446                 $headlineCount = 0;
3447                 $sectionCount = 0; # headlineCount excluding template sections
3448
3449                 # Ugh .. the TOC should have neat indentation levels which can be
3450                 # passed to the skin functions. These are determined here
3451                 $toc = '';
3452                 $full = '';
3453                 $head = array();
3454                 $sublevelCount = array();
3455                 $levelCount = array();
3456                 $toclevel = 0;
3457                 $level = 0;
3458                 $prevlevel = 0;
3459                 $toclevel = 0;
3460                 $prevtoclevel = 0;
3461
3462                 foreach( $matches[3] as $headline ) {
3463                         $istemplate = 0;
3464                         $templatetitle = '';
3465                         $templatesection = 0;
3466                         $numbering = '';
3467                         $mat = array();
3468                         if (preg_match("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", $headline, $mat)) {
3469                                 $istemplate = 1;
3470                                 $templatetitle = base64_decode($mat[1]);
3471                                 $templatesection = 1 + (int)base64_decode($mat[2]);
3472                                 $headline = preg_replace("/<!--MWTEMPLATESECTION=([^&]+)&([^_]+)-->/", "", $headline);
3473                         }
3474
3475                         if( $toclevel ) {
3476                                 $prevlevel = $level;
3477                                 $prevtoclevel = $toclevel;
3478                         }
3479                         $level = $matches[1][$headlineCount];
3480
3481                         if( $doNumberHeadings || $enoughToc ) {
3482
3483                                 if ( $level > $prevlevel ) {
3484                                         # Increase TOC level
3485                                         $toclevel++;
3486                                         $sublevelCount[$toclevel] = 0;
3487                                         if( $toclevel<$wgMaxTocLevel ) {
3488                                                 $toc .= $sk->tocIndent();
3489                                         }
3490                                 }
3491                                 elseif ( $level < $prevlevel && $toclevel > 1 ) {
3492                                         # Decrease TOC level, find level to jump to
3493
3494                                         if ( $toclevel == 2 && $level <= $levelCount[1] ) {
3495                                                 # Can only go down to level 1
3496                                                 $toclevel = 1;
3497                                         } else {
3498                                                 for ($i = $toclevel; $i > 0; $i--) {
3499                                                         if ( $levelCount[$i] == $level ) {
3500                                                                 # Found last matching level
3501                                                                 $toclevel = $i;
3502                                                                 break;
3503                                                         }
3504                                                         elseif ( $levelCount[$i] < $level ) {
3505                                                                 # Found first matching level below current level
3506                                                                 $toclevel = $i + 1;
3507                                                                 break;
3508                                                         }
3509                                                 }
3510                                         }
3511                                         if( $toclevel<$wgMaxTocLevel ) {
3512                                                 $toc .= $sk->tocUnindent( $prevtoclevel - $toclevel );
3513                                         }
3514                                 }
3515                                 else {
3516                                         # No change in level, end TOC line
3517                                         if( $toclevel<$wgMaxTocLevel ) {
3518                                                 $toc .= $sk->tocLineEnd();
3519                                         }
3520                                 }
3521
3522                                 $levelCount[$toclevel] = $level;
3523
3524                                 # count number of headlines for each level
3525                                 @$sublevelCount[$toclevel]++;
3526                                 $dot = 0;
3527                                 for( $i = 1; $i <= $toclevel; $i++ ) {
3528                                         if( !empty( $sublevelCount[$i] ) ) {
3529                                                 if( $dot ) {
3530                                                         $numbering .= '.';
3531                                                 }
3532                                                 $numbering .= $wgContLang->formatNum( $sublevelCount[$i] );
3533                                                 $dot = 1;
3534                                         }
3535                                 }
3536                         }
3537
3538                         # The canonized header is a version of the header text safe to use for links
3539                         # Avoid insertion of weird stuff like <math> by expanding the relevant sections
3540                         $canonized_headline = $this->mStripState->unstripBoth( $headline );
3541
3542                         # Remove link placeholders by the link text.
3543                         #     <!--LINK number-->
3544                         # turns into
3545                         #     link text with suffix
3546                         $canonized_headline = preg_replace( '/<!--LINK ([0-9]*)-->/e',
3547                                                             "\$this->mLinkHolders['texts'][\$1]",
3548                                                             $canonized_headline );
3549                         $canonized_headline = preg_replace( '/<!--IWLINK ([0-9]*)-->/e',
3550                                                             "\$this->mInterwikiLinkHolders['texts'][\$1]",
3551                                                             $canonized_headline );
3552
3553                         # strip out HTML
3554                         $canonized_headline = preg_replace( '/<.*?' . '>/','',$canonized_headline );
3555                         $tocline = trim( $canonized_headline );
3556                         # Save headline for section edit hint before it's escaped
3557                         $headline_hint = trim( $canonized_headline );
3558                         $canonized_headline = Sanitizer::escapeId( $tocline );
3559                         $refers[$headlineCount] = $canonized_headline;
3560
3561                         # count how many in assoc. array so we can track dupes in anchors
3562                         isset( $refers[$canonized_headline] ) ? $refers[$canonized_headline]++ : $refers[$canonized_headline] = 1;
3563                         $refcount[$headlineCount]=$refers[$canonized_headline];
3564
3565                         # Don't number the heading if it is the only one (looks silly)
3566                         if( $doNumberHeadings && count( $matches[3] ) > 1) {
3567                                 # the two are different if the line contains a link
3568                                 $headline=$numbering . ' ' . $headline;
3569                         }
3570
3571                         # Create the anchor for linking from the TOC to the section
3572                         $anchor = $canonized_headline;
3573                         if($refcount[$headlineCount] > 1 ) {
3574                                 $anchor .= '_' . $refcount[$headlineCount];
3575                         }
3576                         if( $enoughToc && ( !isset($wgMaxTocLevel) || $toclevel<$wgMaxTocLevel ) ) {
3577                                 $toc .= $sk->tocLine($anchor, $tocline, $numbering, $toclevel);
3578                         }
3579                         # give headline the correct <h#> tag
3580                         if( $showEditLink && ( !$istemplate || $templatetitle !== "" ) ) {
3581                                 if( $istemplate )
3582                                         $editlink = $sk->editSectionLinkForOther($templatetitle, $templatesection);
3583                                 else
3584                                         $editlink = $sk->editSectionLink($this->mTitle, $sectionCount+1, $headline_hint);
3585                         } else {
3586                                 $editlink = '';
3587                         }
3588                         $head[$headlineCount] = $sk->makeHeadline( $level, $matches['attrib'][$headlineCount], $anchor, $headline, $editlink );
3589
3590                         $headlineCount++;
3591                         if( !$istemplate )
3592                                 $sectionCount++;
3593                 }
3594
3595                 if( $enoughToc ) {
3596                         if( $toclevel<$wgMaxTocLevel ) {
3597                                 $toc .= $sk->tocUnindent( $toclevel - 1 );
3598                         }
3599                         $toc = $sk->tocList( $toc );
3600                 }
3601
3602                 # split up and insert constructed headlines
3603
3604                 $blocks = preg_split( '/<H[1-6].*?' . '>.*?<\/H[1-6]>/i', $text );
3605                 $i = 0;
3606
3607                 foreach( $blocks as $block ) {
3608                         if( $showEditLink && $headlineCount > 0 && $i == 0 && $block != "\n" ) {
3609                                 # This is the [edit] link that appears for the top block of text when
3610                                 # section editing is enabled
3611
3612                                 # Disabled because it broke block formatting
3613                                 # For example, a bullet point in the top line
3614                                 # $full .= $sk->editSectionLink(0);
3615                         }
3616                         $full .= $block;
3617                         if( $enoughToc && !$i && $isMain && !$this->mForceTocPosition ) {
3618                                 # Top anchor now in skin
3619                                 $full = $full.$toc;
3620                         }
3621
3622                         if( !empty( $head[$i] ) ) {
3623                                 $full .= $head[$i];
3624                         }
3625                         $i++;
3626                 }
3627                 if( $this->mForceTocPosition ) {
3628                         return str_replace( '<!--MWTOC-->', $toc, $full );
3629                 } else {
3630                         return $full;
3631                 }
3632         }
3633
3634         /**
3635          * Transform wiki markup when saving a page by doing \r\n -> \n
3636          * conversion, substitting signatures, {{subst:}} templates, etc.
3637          *
3638          * @param string $text the text to transform
3639          * @param Title &$title the Title object for the current article
3640          * @param User &$user the User object describing the current user
3641          * @param ParserOptions $options parsing options
3642          * @param bool $clearState whether to clear the parser state first
3643          * @return string the altered wiki markup
3644          * @public
3645          */
3646         function preSaveTransform( $text, &$title, $user, $options, $clearState = true ) {
3647                 $this->mOptions = $options;
3648                 $this->mTitle =& $title;
3649                 $this->setOutputType( OT_WIKI );
3650
3651                 if ( $clearState ) {
3652                         $this->clearState();
3653                 }
3654
3655                 $stripState = new StripState;
3656                 $pairs = array(
3657                         "\r\n" => "\n",
3658                 );
3659                 $text = str_replace( array_keys( $pairs ), array_values( $pairs ), $text );
3660                 $text = $this->strip( $text, $stripState, true, array( 'gallery' ) );
3661                 $text = $this->pstPass2( $text, $stripState, $user );
3662                 $text = $stripState->unstripBoth( $text );
3663                 return $text;
3664         }
3665
3666         /**
3667          * Pre-save transform helper function
3668          * @private
3669          */
3670         function pstPass2( $text, &$stripState, $user ) {
3671                 global $wgContLang, $wgLocaltimezone;
3672
3673                 /* Note: This is the timestamp saved as hardcoded wikitext to
3674                  * the database, we use $wgContLang here in order to give
3675                  * everyone the same signature and use the default one rather
3676                  * than the one selected in each user's preferences.
3677                  */
3678                 if ( isset( $wgLocaltimezone ) ) {
3679                         $oldtz = getenv( 'TZ' );
3680                         putenv( 'TZ='.$wgLocaltimezone );
3681                 }
3682                 $d = $wgContLang->timeanddate( date( 'YmdHis' ), false, false) .
3683                   ' (' . date( 'T' ) . ')';
3684                 if ( isset( $wgLocaltimezone ) ) {
3685                         putenv( 'TZ='.$oldtz );
3686                 }
3687
3688                 # Variable replacement
3689                 # Because mOutputType is OT_WIKI, this will only process {{subst:xxx}} type tags
3690                 $text = $this->replaceVariables( $text );
3691
3692                 # Strip out <nowiki> etc. added via replaceVariables
3693                 $text = $this->strip( $text, $stripState, false, array( 'gallery' ) );
3694
3695                 # Signatures
3696                 $sigText = $this->getUserSig( $user );
3697                 $text = strtr( $text, array(
3698                         '~~~~~' => $d,
3699                         '~~~~' => "$sigText $d",
3700                         '~~~' => $sigText
3701                 ) );
3702
3703                 # Context links: [[|name]] and [[name (context)|]]
3704                 #
3705                 global $wgLegalTitleChars;
3706                 $tc = "[$wgLegalTitleChars]";
3707                 $nc = '[ _0-9A-Za-z\x80-\xff]'; # Namespaces can use non-ascii!
3708
3709                 $p1 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\))\\|]]/";            # [[ns:page (context)|]]
3710                 $p3 = "/\[\[(:?$nc+:|:|)($tc+?)( \\($tc+\\)|)(, $tc+|)\\|]]/";  # [[ns:page (context), context|]]
3711                 $p2 = "/\[\[\\|($tc+)]]/";                                      # [[|page]]
3712
3713                 # try $p1 first, to turn "[[A, B (C)|]]" into "[[A, B (C)|A, B]]"
3714                 $text = preg_replace( $p1, '[[\\1\\2\\3|\\2]]', $text );
3715                 $text = preg_replace( $p3, '[[\\1\\2\\3\\4|\\2]]', $text );
3716
3717                 $t = $this->mTitle->getText();
3718                 $m = array();
3719                 if ( preg_match( "/^($nc+:|)$tc+?( \\($tc+\\))$/", $t, $m ) ) {
3720                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3721                 } elseif ( preg_match( "/^($nc+:|)$tc+?(, $tc+|)$/", $t, $m ) && '' != "$m[1]$m[2]" ) {
3722                         $text = preg_replace( $p2, "[[$m[1]\\1$m[2]|\\1]]", $text );
3723                 } else {
3724                         # if there's no context, don't bother duplicating the title
3725                         $text = preg_replace( $p2, '[[\\1]]', $text );
3726                 }
3727
3728                 # Trim trailing whitespace
3729                 $text = rtrim( $text );
3730
3731                 return $text;
3732         }
3733
3734         /**
3735          * Fetch the user's signature text, if any, and normalize to
3736          * validated, ready-to-insert wikitext.
3737          *
3738          * @param User $user
3739          * @return string
3740          * @private
3741          */
3742         function getUserSig( &$user ) {
3743                 $username = $user->getName();
3744                 $nickname = $user->getOption( 'nickname' );
3745                 $nickname = $nickname === '' ? $username : $nickname;
3746
3747                 if( $user->getBoolOption( 'fancysig' ) !== false ) {
3748                         # Sig. might contain markup; validate this
3749                         if( $this->validateSig( $nickname ) !== false ) {
3750                                 # Validated; clean up (if needed) and return it
3751                                 return $this->cleanSig( $nickname, true );
3752                         } else {
3753                                 # Failed to validate; fall back to the default
3754                                 $nickname = $username;
3755                                 wfDebug( "Parser::getUserSig: $username has bad XML tags in signature.\n" );
3756                         }
3757                 }
3758
3759                 // Make sure nickname doesnt get a sig in a sig
3760                 $nickname = $this->cleanSigInSig( $nickname );
3761
3762                 # If we're still here, make it a link to the user page
3763                 $userpage = $user->getUserPage();
3764                 return( '[[' . $userpage->getPrefixedText() . '|' . wfEscapeWikiText( $nickname ) . ']]' );
3765         }
3766
3767         /**
3768          * Check that the user's signature contains no bad XML
3769          *
3770          * @param string $text
3771          * @return mixed An expanded string, or false if invalid.
3772          */
3773         function validateSig( $text ) {
3774                 return( wfIsWellFormedXmlFragment( $text ) ? $text : false );
3775         }
3776
3777         /**
3778          * Clean up signature text
3779          *
3780          * 1) Strip ~~~, ~~~~ and ~~~~~ out of signatures @see cleanSigInSig
3781          * 2) Substitute all transclusions
3782          *
3783          * @param string $text
3784          * @param $parsing Whether we're cleaning (preferences save) or parsing
3785          * @return string Signature text
3786          */
3787         function cleanSig( $text, $parsing = false ) {
3788                 global $wgTitle;
3789                 $this->startExternalParse( $wgTitle, new ParserOptions(), $parsing ? OT_WIKI : OT_MSG );
3790
3791                 $substWord = MagicWord::get( 'subst' );
3792                 $substRegex = '/\{\{(?!(?:' . $substWord->getBaseRegex() . '))/x' . $substWord->getRegexCase();
3793                 $substText = '{{' . $substWord->getSynonym( 0 );
3794
3795                 $text = preg_replace( $substRegex, $substText, $text );
3796                 $text = $this->cleanSigInSig( $text );
3797                 $text = $this->replaceVariables( $text );
3798
3799                 $this->clearState();
3800                 return $text;
3801         }
3802
3803         /**
3804          * Strip ~~~, ~~~~ and ~~~~~ out of signatures
3805          * @param string $text
3806          * @return string Signature text with /~{3,5}/ removed
3807          */
3808         function cleanSigInSig( $text ) {
3809                 $text = preg_replace( '/~{3,5}/', '', $text );
3810                 return $text;
3811         }
3812
3813         /**
3814          * Set up some variables which are usually set up in parse()
3815          * so that an external function can call some class members with confidence
3816          * @public
3817          */
3818         function startExternalParse( &$title, $options, $outputType, $clearState = true ) {
3819                 $this->mTitle =& $title;
3820                 $this->mOptions = $options;
3821                 $this->setOutputType( $outputType );
3822                 if ( $clearState ) {
3823                         $this->clearState();
3824                 }
3825         }
3826
3827         /**
3828          * Transform a MediaWiki message by replacing magic variables.
3829          *
3830          * @param string $text the text to transform
3831          * @param ParserOptions $options  options
3832          * @return string the text with variables substituted
3833          * @public
3834          */
3835         function transformMsg( $text, $options ) {
3836                 global $wgTitle;
3837                 static $executing = false;
3838
3839                 $fname = "Parser::transformMsg";
3840
3841                 # Guard against infinite recursion
3842                 if ( $executing ) {
3843                         return $text;
3844                 }
3845                 $executing = true;
3846
3847                 wfProfileIn($fname);
3848
3849                 if ( $wgTitle && !( $wgTitle instanceof FakeTitle ) ) {
3850                         $this->mTitle = $wgTitle;
3851                 } else {
3852                         $this->mTitle = Title::newFromText('msg');
3853                 }
3854                 $this->mOptions = $options;
3855                 $this->setOutputType( OT_MSG );
3856                 $this->clearState();
3857                 $text = $this->replaceVariables( $text );
3858
3859                 $executing = false;
3860                 wfProfileOut($fname);
3861                 return $text;
3862         }
3863
3864         /**
3865          * Create an HTML-style tag, e.g. <yourtag>special text</yourtag>
3866          * The callback should have the following form:
3867          *    function myParserHook( $text, $params, &$parser ) { ... }
3868          *
3869          * Transform and return $text. Use $parser for any required context, e.g. use
3870          * $parser->getTitle() and $parser->getOptions() not $wgTitle or $wgOut->mParserOptions
3871          *
3872          * @public
3873          *
3874          * @param mixed $tag The tag to use, e.g. 'hook' for <hook>
3875          * @param mixed $callback The callback function (and object) to use for the tag
3876          *
3877          * @return The old value of the mTagHooks array associated with the hook
3878          */
3879         function setHook( $tag, $callback ) {
3880                 $tag = strtolower( $tag );
3881                 $oldVal = isset( $this->mTagHooks[$tag] ) ? $this->mTagHooks[$tag] : null;
3882                 $this->mTagHooks[$tag] = $callback;
3883
3884                 return $oldVal;
3885         }
3886
3887         /**
3888          * Create a function, e.g. {{sum:1|2|3}}
3889          * The callback function should have the form:
3890          *    function myParserFunction( &$parser, $arg1, $arg2, $arg3 ) { ... }
3891          *
3892          * The callback may either return the text result of the function, or an array with the text
3893          * in element 0, and a number of flags in the other elements. The names of the flags are
3894          * specified in the keys. Valid flags are:
3895          *   found                     The text returned is valid, stop processing the template. This
3896          *                             is on by default.
3897          *   nowiki                    Wiki markup in the return value should be escaped
3898          *   noparse                   Unsafe HTML tags should not be stripped, etc.
3899          *   noargs                    Don't replace triple-brace arguments in the return value
3900          *   isHTML                    The returned text is HTML, armour it against wikitext transformation
3901          *
3902          * @public
3903          *
3904          * @param string $id The magic word ID
3905          * @param mixed $callback The callback function (and object) to use
3906          * @param integer $flags a combination of the following flags:
3907          *                SFH_NO_HASH No leading hash, i.e. {{plural:...}} instead of {{#if:...}}
3908          *
3909          * @return The old callback function for this name, if any
3910          */
3911         function setFunctionHook( $id, $callback, $flags = 0 ) {
3912                 $oldVal = isset( $this->mFunctionHooks[$id] ) ? $this->mFunctionHooks[$id] : null;
3913                 $this->mFunctionHooks[$id] = $callback;
3914
3915                 # Add to function cache
3916                 $mw = MagicWord::get( $id );
3917                 if( !$mw )
3918                         throw new MWException( 'Parser::setFunctionHook() expecting a magic word identifier.' );
3919
3920                 $synonyms = $mw->getSynonyms();
3921                 $sensitive = intval( $mw->isCaseSensitive() );
3922
3923                 foreach ( $synonyms as $syn ) {
3924                         # Case
3925                         if ( !$sensitive ) {
3926                                 $syn = strtolower( $syn );
3927                         }
3928                         # Add leading hash
3929                         if ( !( $flags & SFH_NO_HASH ) ) {
3930                                 $syn = '#' . $syn;
3931                         }
3932                         # Remove trailing colon
3933                         if ( substr( $syn, -1, 1 ) == ':' ) {
3934                                 $syn = substr( $syn, 0, -1 );
3935                         }
3936                         $this->mFunctionSynonyms[$sensitive][$syn] = $id;
3937                 }
3938                 return $oldVal;
3939         }
3940
3941         /**
3942          * Get all registered function hook identifiers
3943          *
3944          * @return array
3945          */
3946         function getFunctionHooks() {
3947                 return array_keys( $this->mFunctionHooks );
3948         }
3949
3950         /**
3951          * Replace <!--LINK--> link placeholders with actual links, in the buffer
3952          * Placeholders created in Skin::makeLinkObj()
3953          * Returns an array of links found, indexed by PDBK:
3954          *  0 - broken
3955          *  1 - normal link
3956          *  2 - stub
3957          * $options is a bit field, RLH_FOR_UPDATE to select for update
3958          */
3959         function replaceLinkHolders( &$text, $options = 0 ) {
3960                 global $wgUser;
3961                 global $wgContLang;
3962
3963                 $fname = 'Parser::replaceLinkHolders';
3964                 wfProfileIn( $fname );
3965
3966                 $pdbks = array();
3967                 $colours = array();
3968                 $sk = $this->mOptions->getSkin();
3969                 $linkCache =& LinkCache::singleton();
3970
3971                 if ( !empty( $this->mLinkHolders['namespaces'] ) ) {
3972                         wfProfileIn( $fname.'-check' );
3973                         $dbr = wfGetDB( DB_SLAVE );
3974                         $page = $dbr->tableName( 'page' );
3975                         $threshold = $wgUser->getOption('stubthreshold');
3976
3977                         # Sort by namespace
3978                         asort( $this->mLinkHolders['namespaces'] );
3979
3980                         # Generate query
3981                         $query = false;
3982                         $current = null;
3983                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
3984                                 # Make title object
3985                                 $title = $this->mLinkHolders['titles'][$key];
3986
3987                                 # Skip invalid entries.
3988                                 # Result will be ugly, but prevents crash.
3989                                 if ( is_null( $title ) ) {
3990                                         continue;
3991                                 }
3992                                 $pdbk = $pdbks[$key] = $title->getPrefixedDBkey();
3993
3994                                 # Check if it's a static known link, e.g. interwiki
3995                                 if ( $title->isAlwaysKnown() ) {
3996                                         $colours[$pdbk] = 1;
3997                                 } elseif ( ( $id = $linkCache->getGoodLinkID( $pdbk ) ) != 0 ) {
3998                                         $colours[$pdbk] = 1;
3999                                         $this->mOutput->addLink( $title, $id );
4000                                 } elseif ( $linkCache->isBadLink( $pdbk ) ) {
4001                                         $colours[$pdbk] = 0;
4002                                 } else {
4003                                         # Not in the link cache, add it to the query
4004                                         if ( !isset( $current ) ) {
4005                                                 $current = $ns;
4006                                                 $query =  "SELECT page_id, page_namespace, page_title";
4007                                                 if ( $threshold > 0 ) {
4008                                                         $query .= ', page_len, page_is_redirect';
4009                                                 }
4010                                                 $query .= " FROM $page WHERE (page_namespace=$ns AND page_title IN(";
4011                                         } elseif ( $current != $ns ) {
4012                                                 $current = $ns;
4013                                                 $query .= ")) OR (page_namespace=$ns AND page_title IN(";
4014                                         } else {
4015                                                 $query .= ', ';
4016                                         }
4017
4018                                         $query .= $dbr->addQuotes( $this->mLinkHolders['dbkeys'][$key] );
4019                                 }
4020                         }
4021                         if ( $query ) {
4022                                 $query .= '))';
4023                                 if ( $options & RLH_FOR_UPDATE ) {
4024                                         $query .= ' FOR UPDATE';
4025                                 }
4026
4027                                 $res = $dbr->query( $query, $fname );
4028
4029                                 # Fetch data and form into an associative array
4030                                 # non-existent = broken
4031                                 # 1 = known
4032                                 # 2 = stub
4033                                 while ( $s = $dbr->fetchObject($res) ) {
4034                                         $title = Title::makeTitle( $s->page_namespace, $s->page_title );
4035                                         $pdbk = $title->getPrefixedDBkey();
4036                                         $linkCache->addGoodLinkObj( $s->page_id, $title );
4037                                         $this->mOutput->addLink( $title, $s->page_id );
4038
4039                                         if ( $threshold >  0 ) {
4040                                                 $size = $s->page_len;
4041                                                 if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4042                                                         $colours[$pdbk] = 1;
4043                                                 } else {
4044                                                         $colours[$pdbk] = 2;
4045                                                 }
4046                                         } else {
4047                                                 $colours[$pdbk] = 1;
4048                                         }
4049                                 }
4050                         }
4051                         wfProfileOut( $fname.'-check' );
4052
4053                         # Do a second query for different language variants of links and categories
4054                         if($wgContLang->hasVariants()){
4055                                 $linkBatch = new LinkBatch();
4056                                 $variantMap = array(); // maps $pdbkey_Variant => $keys (of link holders)
4057                                 $categoryMap = array(); // maps $category_variant => $category (dbkeys)
4058                                 $varCategories = array(); // category replacements oldDBkey => newDBkey
4059
4060                                 $categories = $this->mOutput->getCategoryLinks();
4061
4062                                 // Add variants of links to link batch
4063                                 foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4064                                         $title = $this->mLinkHolders['titles'][$key];
4065                                         if ( is_null( $title ) )
4066                                                 continue;
4067
4068                                         $pdbk = $title->getPrefixedDBkey();
4069                                         $titleText = $title->getText();
4070
4071                                         // generate all variants of the link title text
4072                                         $allTextVariants = $wgContLang->convertLinkToAllVariants($titleText);
4073
4074                                         // if link was not found (in first query), add all variants to query
4075                                         if ( !isset($colours[$pdbk]) ){
4076                                                 foreach($allTextVariants as $textVariant){
4077                                                         if($textVariant != $titleText){
4078                                                                 $variantTitle = Title::makeTitle( $ns, $textVariant );
4079                                                                 if(is_null($variantTitle)) continue;
4080                                                                 $linkBatch->addObj( $variantTitle );
4081                                                                 $variantMap[$variantTitle->getPrefixedDBkey()][] = $key;
4082                                                         }
4083                                                 }
4084                                         }
4085                                 }
4086
4087                                 // process categories, check if a category exists in some variant
4088                                 foreach( $categories as $category){
4089                                         $variants = $wgContLang->convertLinkToAllVariants($category);
4090                                         foreach($variants as $variant){
4091                                                 if($variant != $category){
4092                                                         $variantTitle = Title::newFromDBkey( Title::makeName(NS_CATEGORY,$variant) );
4093                                                         if(is_null($variantTitle)) continue;
4094                                                         $linkBatch->addObj( $variantTitle );
4095                                                         $categoryMap[$variant] = $category;
4096                                                 }
4097                                         }
4098                                 }
4099
4100
4101                                 if(!$linkBatch->isEmpty()){
4102                                         // construct query
4103                                         $titleClause = $linkBatch->constructSet('page', $dbr);
4104
4105                                         $variantQuery =  "SELECT page_id, page_namespace, page_title";
4106                                         if ( $threshold > 0 ) {
4107                                                 $variantQuery .= ', page_len, page_is_redirect';
4108                                         }
4109
4110                                         $variantQuery .= " FROM $page WHERE $titleClause";
4111                                         if ( $options & RLH_FOR_UPDATE ) {
4112                                                 $variantQuery .= ' FOR UPDATE';
4113                                         }
4114
4115                                         $varRes = $dbr->query( $variantQuery, $fname );
4116
4117                                         // for each found variants, figure out link holders and replace
4118                                         while ( $s = $dbr->fetchObject($varRes) ) {
4119
4120                                                 $variantTitle = Title::makeTitle( $s->page_namespace, $s->page_title );
4121                                                 $varPdbk = $variantTitle->getPrefixedDBkey();
4122                                                 $vardbk = $variantTitle->getDBkey();
4123
4124                                                 $holderKeys = array();
4125                                                 if(isset($variantMap[$varPdbk])){
4126                                                         $holderKeys = $variantMap[$varPdbk];
4127                                                         $linkCache->addGoodLinkObj( $s->page_id, $variantTitle );
4128                                                         $this->mOutput->addLink( $variantTitle, $s->page_id );
4129                                                 }
4130
4131                                                 // loop over link holders
4132                                                 foreach($holderKeys as $key){
4133                                                         $title = $this->mLinkHolders['titles'][$key];
4134                                                         if ( is_null( $title ) ) continue;
4135
4136                                                         $pdbk = $title->getPrefixedDBkey();
4137
4138                                                         if(!isset($colours[$pdbk])){
4139                                                                 // found link in some of the variants, replace the link holder data
4140                                                                 $this->mLinkHolders['titles'][$key] = $variantTitle;
4141                                                                 $this->mLinkHolders['dbkeys'][$key] = $variantTitle->getDBkey();
4142
4143                                                                 // set pdbk and colour
4144                                                                 $pdbks[$key] = $varPdbk;
4145                                                                 if ( $threshold >  0 ) {
4146                                                                         $size = $s->page_len;
4147                                                                         if ( $s->page_is_redirect || $s->page_namespace != 0 || $size >= $threshold ) {
4148                                                                                 $colours[$varPdbk] = 1;
4149                                                                         } else {
4150                                                                                 $colours[$varPdbk] = 2;
4151                                                                         }
4152                                                                 }
4153                                                                 else {
4154                                                                         $colours[$varPdbk] = 1;
4155                                                                 }
4156                                                         }
4157                                                 }
4158
4159                                                 // check if the object is a variant of a category
4160                                                 if(isset($categoryMap[$vardbk])){
4161                                                         $oldkey = $categoryMap[$vardbk];
4162                                                         if($oldkey != $vardbk)
4163                                                                 $varCategories[$oldkey]=$vardbk;
4164                                                 }
4165                                         }
4166
4167                                         // rebuild the categories in original order (if there are replacements)
4168                                         if(count($varCategories)>0){
4169                                                 $newCats = array();
4170                                                 $originalCats = $this->mOutput->getCategories();
4171                                                 foreach($originalCats as $cat => $sortkey){
4172                                                         // make the replacement
4173                                                         if( array_key_exists($cat,$varCategories) )
4174                                                                 $newCats[$varCategories[$cat]] = $sortkey;
4175                                                         else $newCats[$cat] = $sortkey;
4176                                                 }
4177                                                 $this->mOutput->setCategoryLinks($newCats);
4178                                         }
4179                                 }
4180                         }
4181
4182                         # Construct search and replace arrays
4183                         wfProfileIn( $fname.'-construct' );
4184                         $replacePairs = array();
4185                         foreach ( $this->mLinkHolders['namespaces'] as $key => $ns ) {
4186                                 $pdbk = $pdbks[$key];
4187                                 $searchkey = "<!--LINK $key-->";
4188                                 $title = $this->mLinkHolders['titles'][$key];
4189                                 if ( empty( $colours[$pdbk] ) ) {
4190                                         $linkCache->addBadLinkObj( $title );
4191                                         $colours[$pdbk] = 0;
4192                                         $this->mOutput->addLink( $title, 0 );
4193                                         $replacePairs[$searchkey] = $sk->makeBrokenLinkObj( $title,
4194                                                                         $this->mLinkHolders['texts'][$key],
4195                                                                         $this->mLinkHolders['queries'][$key] );
4196                                 } elseif ( $colours[$pdbk] == 1 ) {
4197                                         $replacePairs[$searchkey] = $sk->makeKnownLinkObj( $title,
4198                                                                         $this->mLinkHolders['texts'][$key],
4199                                                                         $this->mLinkHolders['queries'][$key] );
4200                                 } elseif ( $colours[$pdbk] == 2 ) {
4201                                         $replacePairs[$searchkey] = $sk->makeStubLinkObj( $title,
4202                                                                         $this->mLinkHolders['texts'][$key],
4203                                                                         $this->mLinkHolders['queries'][$key] );
4204                                 }
4205                         }
4206                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4207                         wfProfileOut( $fname.'-construct' );
4208
4209                         # Do the thing
4210                         wfProfileIn( $fname.'-replace' );
4211                         $text = preg_replace_callback(
4212                                 '/(<!--LINK .*?-->)/',
4213                                 $replacer->cb(),
4214                                 $text);
4215
4216                         wfProfileOut( $fname.'-replace' );
4217                 }
4218
4219                 # Now process interwiki link holders
4220                 # This is quite a bit simpler than internal links
4221                 if ( !empty( $this->mInterwikiLinkHolders['texts'] ) ) {
4222                         wfProfileIn( $fname.'-interwiki' );
4223                         # Make interwiki link HTML
4224                         $replacePairs = array();
4225                         foreach( $this->mInterwikiLinkHolders['texts'] as $key => $link ) {
4226                                 $title = $this->mInterwikiLinkHolders['titles'][$key];
4227                                 $replacePairs[$key] = $sk->makeLinkObj( $title, $link );
4228                         }
4229                         $replacer = new HashtableReplacer( $replacePairs, 1 );
4230
4231                         $text = preg_replace_callback(
4232                                 '/<!--IWLINK (.*?)-->/',
4233                                 $replacer->cb(),
4234                                 $text );
4235                         wfProfileOut( $fname.'-interwiki' );
4236                 }
4237
4238                 wfProfileOut( $fname );
4239                 return $colours;
4240         }
4241
4242         /**
4243          * Replace <!--LINK--> link placeholders with plain text of links
4244          * (not HTML-formatted).
4245          * @param string $text
4246          * @return string
4247          */
4248         function replaceLinkHoldersText( $text ) {
4249                 $fname = 'Parser::replaceLinkHoldersText';
4250                 wfProfileIn( $fname );
4251
4252                 $text = preg_replace_callback(
4253                         '/<!--(LINK|IWLINK) (.*?)-->/',
4254                         array( &$this, 'replaceLinkHoldersTextCallback' ),
4255                         $text );
4256
4257                 wfProfileOut( $fname );
4258                 return $text;
4259         }
4260
4261         /**
4262          * @param array $matches
4263          * @return string
4264          * @private
4265          */
4266         function replaceLinkHoldersTextCallback( $matches ) {
4267                 $type = $matches[1];
4268                 $key  = $matches[2];
4269                 if( $type == 'LINK' ) {
4270                         if( isset( $this->mLinkHolders['texts'][$key] ) ) {
4271                                 return $this->mLinkHolders['texts'][$key];
4272                         }
4273                 } elseif( $type == 'IWLINK' ) {
4274                         if( isset( $this->mInterwikiLinkHolders['texts'][$key] ) ) {
4275                                 return $this->mInterwikiLinkHolders['texts'][$key];
4276                         }
4277                 }
4278                 return $matches[0];
4279         }
4280
4281         /**
4282          * Tag hook handler for 'pre'.
4283          */
4284         function renderPreTag( $text, $attribs ) {
4285                 // Backwards-compatibility hack
4286                 $content = StringUtils::delimiterReplace( '<nowiki>', '</nowiki>', '$1', $text, 'i' );
4287
4288                 $attribs = Sanitizer::validateTagAttributes( $attribs, 'pre' );
4289                 return wfOpenElement( 'pre', $attribs ) .
4290                         Xml::escapeTagsOnly( $content ) .
4291                         '</pre>';
4292         }
4293
4294         /**
4295          * Renders an image gallery from a text with one line per image.
4296          * text labels may be given by using |-style alternative text. E.g.
4297          *   Image:one.jpg|The number "1"
4298          *   Image:tree.jpg|A tree
4299          * given as text will return the HTML of a gallery with two images,
4300          * labeled 'The number "1"' and
4301          * 'A tree'.
4302          */
4303         function renderImageGallery( $text, $params ) {
4304                 $ig = new ImageGallery();
4305                 $ig->setContextTitle( $this->mTitle );
4306                 $ig->setShowBytes( false );
4307                 $ig->setShowFilename( false );
4308                 $ig->setParsing();
4309                 $ig->useSkin( $this->mOptions->getSkin() );
4310
4311                 if( isset( $params['caption'] ) ) {
4312                         $caption = $params['caption'];
4313                         $caption = htmlspecialchars( $caption );
4314                         $caption = $this->replaceInternalLinks( $caption );
4315                         $ig->setCaptionHtml( $caption );
4316                 }
4317                 if( isset( $params['perrow'] ) ) {
4318                         $ig->setPerRow( $params['perrow'] );
4319                 }
4320                 if( isset( $params['widths'] ) ) {
4321                         $ig->setWidths( $params['widths'] );
4322                 }
4323                 if( isset( $params['heights'] ) ) {
4324                         $ig->setHeights( $params['heights'] );
4325                 }
4326
4327                 $lines = explode( "\n", $text );
4328                 foreach ( $lines as $line ) {
4329                         # match lines like these:
4330                         # Image:someimage.jpg|This is some image
4331                         $matches = array();
4332                         preg_match( "/^([^|]+)(\\|(.*))?$/", $line, $matches );
4333                         # Skip empty lines
4334                         if ( count( $matches ) == 0 ) {
4335                                 continue;
4336                         }
4337                         $tp = Title::newFromText( $matches[1] );
4338                         $nt =& $tp;
4339                         if( is_null( $nt ) ) {
4340                                 # Bogus title. Ignore these so we don't bomb out later.
4341                                 continue;
4342                         }
4343                         if ( isset( $matches[3] ) ) {
4344                                 $label = $matches[3];
4345                         } else {
4346                                 $label = '';
4347                         }
4348
4349                         $pout = $this->parse( $label,
4350                                 $this->mTitle,
4351                                 $this->mOptions,
4352                                 false, // Strip whitespace...?
4353                                 false  // Don't clear state!
4354                         );
4355                         $html = $pout->getText();
4356
4357                         $ig->add( new Image( $nt ), $html );
4358
4359                         # Only add real images (bug #5586)
4360                         if ( $nt->getNamespace() == NS_IMAGE ) {
4361                                 $this->mOutput->addImage( $nt->getDBkey() );
4362                         }
4363                 }
4364                 return $ig->toHTML();
4365         }
4366
4367         /**
4368          * Parse image options text and use it to make an image
4369          */
4370         function makeImage( $nt, $options ) {
4371                 global $wgUseImageResize, $wgDjvuRenderer;
4372
4373                 # Check if the options text is of the form "options|alt text"
4374                 # Options are:
4375                 #  * thumbnail          make a thumbnail with enlarge-icon and caption, alignment depends on lang
4376                 #  * left               no resizing, just left align. label is used for alt= only
4377                 #  * right              same, but right aligned
4378                 #  * none               same, but not aligned
4379                 #  * ___px              scale to ___ pixels width, no aligning. e.g. use in taxobox
4380                 #  * center             center the image
4381                 #  * framed             Keep original image size, no magnify-button.
4382                 # vertical-align values (no % or length right now):
4383                 #  * baseline
4384                 #  * sub
4385                 #  * super
4386                 #  * top
4387                 #  * text-top
4388                 #  * middle
4389                 #  * bottom
4390                 #  * text-bottom
4391
4392                 $part = explode( '|', $options);
4393
4394                 $mwAlign = array();
4395                 $alignments = array( 'left', 'right', 'center', 'none', 'baseline', 'sub', 'super', 'top', 'text-top', 'middle', 'bottom', 'text-bottom' );
4396                 foreach ( $alignments as $alignment ) {
4397                         $mwAlign[$alignment] =& MagicWord::get( 'img_'.$alignment );
4398                 }
4399                 $mwThumb  =& MagicWord::get( 'img_thumbnail' );
4400                 $mwManualThumb =& MagicWord::get( 'img_manualthumb' );
4401                 $mwWidth  =& MagicWord::get( 'img_width' );
4402                 $mwFramed =& MagicWord::get( 'img_framed' );
4403                 $mwPage   =& MagicWord::get( 'img_page' );
4404                 $caption = '';
4405
4406                 $width = $height = $framed = $thumb = false;
4407                 $page = null;
4408                 $manual_thumb = '' ;
4409                 $align = $valign = '';
4410
4411                 foreach( $part as $val ) {
4412                         if ( $wgUseImageResize && ! is_null( $mwThumb->matchVariableStartToEnd($val) ) ) {
4413                                 $thumb=true;
4414                         } elseif ( ! is_null( $match = $mwManualThumb->matchVariableStartToEnd($val) ) ) {
4415                                 # use manually specified thumbnail
4416                                 $thumb=true;
4417                                 $manual_thumb = $match;
4418                         } else {
4419                                 foreach( $alignments as $alignment ) {
4420                                         if ( ! is_null( $mwAlign[$alignment]->matchVariableStartToEnd($val) ) ) {
4421                                                 switch ( $alignment ) {
4422                                                         case 'left': case 'right': case 'center': case 'none':
4423                                                                 $align = $alignment; break;
4424                                                         default:
4425                                                                 $valign = $alignment;
4426                                                 }
4427                                                 continue 2;
4428                                         }
4429                                 }
4430                                 if ( isset( $wgDjvuRenderer ) && $wgDjvuRenderer
4431                                 && ! is_null( $match = $mwPage->matchVariableStartToEnd($val) ) ) {
4432                                         # Select a page in a multipage document
4433                                         $page = $match;
4434                                 } elseif ( $wgUseImageResize && !$width && ! is_null( $match = $mwWidth->matchVariableStartToEnd($val) ) ) {
4435                                         wfDebug( "img_width match: $match\n" );
4436                                         # $match is the image width in pixels
4437                                         $m = array();
4438                                         if ( preg_match( '/^([0-9]*)x([0-9]*)$/', $match, $m ) ) {
4439                                                 $width = intval( $m[1] );
4440                                                 $height = intval( $m[2] );
4441                                         } else {
4442                                                 $width = intval($match);
4443                                         }
4444                                 } elseif ( ! is_null( $mwFramed->matchVariableStartToEnd($val) ) ) {
4445                                         $framed=true;
4446                                 } else {
4447                                         $caption = $val;
4448                                 }
4449                         }
4450                 }
4451                 # Strip bad stuff out of the alt text
4452                 $alt = $this->replaceLinkHoldersText( $caption );
4453
4454                 # make sure there are no placeholders in thumbnail attributes
4455                 # that are later expanded to html- so expand them now and
4456                 # remove the tags
4457                 $alt = $this->mStripState->unstripBoth( $alt );
4458                 $alt = Sanitizer::stripAllTags( $alt );
4459
4460                 # Linker does the rest
4461                 $sk = $this->mOptions->getSkin();
4462                 return $sk->makeImageLinkObj( $nt, $caption, $alt, $align, $width, $height, $framed, $thumb, $manual_thumb, $page, $valign );
4463         }
4464
4465         /**
4466          * Set a flag in the output object indicating that the content is dynamic and
4467          * shouldn't be cached.
4468          */
4469         function disableCache() {
4470                 wfDebug( "Parser output marked as uncacheable.\n" );
4471                 $this->mOutput->mCacheTime = -1;
4472         }
4473
4474         /**#@+
4475          * Callback from the Sanitizer for expanding items found in HTML attribute
4476          * values, so they can be safely tested and escaped.
4477          * @param string $text
4478          * @param array $args
4479          * @return string
4480          * @private
4481          */
4482         function attributeStripCallback( &$text, $args ) {
4483                 $text = $this->replaceVariables( $text, $args );
4484                 $text = $this->mStripState->unstripBoth( $text );
4485                 return $text;
4486         }
4487
4488         /**#@-*/
4489
4490         /**#@+
4491          * Accessor/mutator
4492          */
4493         function Title( $x = NULL ) { return wfSetVar( $this->mTitle, $x ); }
4494         function Options( $x = NULL ) { return wfSetVar( $this->mOptions, $x ); }
4495         function OutputType( $x = NULL ) { return wfSetVar( $this->mOutputType, $x ); }
4496         /**#@-*/
4497
4498         /**#@+
4499          * Accessor
4500          */
4501         function getTags() { return array_keys( $this->mTagHooks ); }
4502         /**#@-*/
4503
4504
4505         /**
4506          * Break wikitext input into sections, and either pull or replace
4507          * some particular section's text.
4508          *
4509          * External callers should use the getSection and replaceSection methods.
4510          *
4511          * @param $text Page wikitext
4512          * @param $section Numbered section. 0 pulls the text before the first
4513          *                 heading; other numbers will pull the given section
4514          *                 along with its lower-level subsections.
4515          * @param $mode One of "get" or "replace"
4516          * @param $newtext Replacement text for section data.
4517          * @return string for "get", the extracted section text.
4518          *                for "replace", the whole page with the section replaced.
4519          */
4520         private function extractSections( $text, $section, $mode, $newtext='' ) {
4521                 # strip NOWIKI etc. to avoid confusion (true-parameter causes HTML
4522                 # comments to be stripped as well)
4523                 $stripState = new StripState;
4524
4525                 $oldOutputType = $this->mOutputType;
4526                 $oldOptions = $this->mOptions;
4527                 $this->mOptions = new ParserOptions();
4528                 $this->setOutputType( OT_WIKI );
4529
4530                 $striptext = $this->strip( $text, $stripState, true );
4531
4532                 $this->setOutputType( $oldOutputType );
4533                 $this->mOptions = $oldOptions;
4534
4535                 # now that we can be sure that no pseudo-sections are in the source,
4536                 # split it up by section
4537                 $uniq = preg_quote( $this->uniqPrefix(), '/' );
4538                 $comment = "(?:$uniq-!--.*?QINU)";
4539                 $secs = preg_split(
4540                 /*
4541                         "/
4542                         ^(
4543                         (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4544                         (?:
4545                                 (=+) # Should this be limited to 6?
4546                                 .+?  # Section title...
4547                                 \\2  # Ending = count must match start
4548                         |
4549                                 ^
4550                                 <h([1-6])\b.*?>
4551                                 .*?
4552                                 <\/h\\3\s*>
4553                         )
4554                         (?:$comment|<\/?noinclude>|\s+)* # Trailing whitespace ok
4555                         )$
4556                         /mix",
4557                 */
4558                         "/
4559                         (
4560                                 ^
4561                                 (?:$comment|<\/?noinclude>)* # Initial comments will be stripped
4562                                 (=+) # Should this be limited to 6?
4563                                 .+?  # Section title...
4564                                 \\2  # Ending = count must match start
4565                                 (?:$comment|<\/?noinclude>|[ \\t]+)* # Trailing whitespace ok
4566                                 $
4567                         |
4568                                 <h([1-6])\b.*?>
4569                                 .*?
4570                                 <\/h\\3\s*>
4571                         )
4572                         /mix",
4573                         $striptext, -1,
4574                         PREG_SPLIT_DELIM_CAPTURE);
4575
4576                 if( $mode == "get" ) {
4577                         if( $section == 0 ) {
4578                                 // "Section 0" returns the content before any other section.
4579                                 $rv = $secs[0];
4580                         } else {
4581                                 $rv = "";
4582                         }
4583                 } elseif( $mode == "replace" ) {
4584                         if( $section == 0 ) {
4585                                 $rv = $newtext . "\n\n";
4586                                 $remainder = true;
4587                         } else {
4588                                 $rv = $secs[0];
4589                                 $remainder = false;
4590                         }
4591                 }
4592                 $count = 0;
4593                 $sectionLevel = 0;
4594                 for( $index = 1; $index < count( $secs ); ) {
4595                         $headerLine = $secs[$index++];
4596                         if( $secs[$index] ) {
4597                                 // A wiki header
4598                                 $headerLevel = strlen( $secs[$index++] );
4599                         } else {
4600                                 // An HTML header
4601                                 $index++;
4602                                 $headerLevel = intval( $secs[$index++] );
4603                         }
4604                         $content = $secs[$index++];
4605
4606                         $count++;
4607                         if( $mode == "get" ) {
4608                                 if( $count == $section ) {
4609                                         $rv = $headerLine . $content;
4610                                         $sectionLevel = $headerLevel;
4611                                 } elseif( $count > $section ) {
4612                                         if( $sectionLevel && $headerLevel > $sectionLevel ) {
4613                                                 $rv .= $headerLine . $content;
4614                                         } else {
4615                                                 // Broke out to a higher-level section
4616                                                 break;
4617                                         }
4618                                 }
4619                         } elseif( $mode == "replace" ) {
4620                                 if( $count < $section ) {
4621                                         $rv .= $headerLine . $content;
4622                                 } elseif( $count == $section ) {
4623                                         $rv .= $newtext . "\n\n";
4624                                         $sectionLevel = $headerLevel;
4625                                 } elseif( $count > $section ) {
4626                                         if( $headerLevel <= $sectionLevel ) {
4627                                                 // Passed the section's sub-parts.
4628                                                 $remainder = true;
4629                                         }
4630                                         if( $remainder ) {
4631                                                 $rv .= $headerLine . $content;
4632                                         }
4633                                 }
4634                         }
4635                 }
4636                 # reinsert stripped tags
4637                 $rv = trim( $stripState->unstripBoth( $rv ) );
4638                 return $rv;
4639         }
4640
4641         /**
4642          * This function returns the text of a section, specified by a number ($section).
4643          * A section is text under a heading like == Heading == or \<h1\>Heading\</h1\>, or
4644          * the first section before any such heading (section 0).
4645          *
4646          * If a section contains subsections, these are also returned.
4647          *
4648          * @param $text String: text to look in
4649          * @param $section Integer: section number
4650          * @return string text of the requested section
4651          */
4652         public function getSection( $text, $section ) {
4653                 return $this->extractSections( $text, $section, "get" );
4654         }
4655
4656         public function replaceSection( $oldtext, $section, $text ) {
4657                 return $this->extractSections( $oldtext, $section, "replace", $text );
4658         }
4659
4660         /**
4661          * Get the timestamp associated with the current revision, adjusted for
4662          * the default server-local timestamp
4663          */
4664         function getRevisionTimestamp() {
4665                 if ( is_null( $this->mRevisionTimestamp ) ) {
4666                         wfProfileIn( __METHOD__ );
4667                         global $wgContLang;
4668                         $dbr = wfGetDB( DB_SLAVE );
4669                         $timestamp = $dbr->selectField( 'revision', 'rev_timestamp',
4670                                         array( 'rev_id' => $this->mRevisionId ), __METHOD__ );
4671
4672                         // Normalize timestamp to internal MW format for timezone processing.
4673                         // This has the added side-effect of replacing a null value with
4674                         // the current time, which gives us more sensible behavior for
4675                         // previews.
4676                         $timestamp = wfTimestamp( TS_MW, $timestamp );
4677
4678                         // The cryptic '' timezone parameter tells to use the site-default
4679                         // timezone offset instead of the user settings.
4680                         //
4681                         // Since this value will be saved into the parser cache, served
4682                         // to other users, and potentially even used inside links and such,
4683                         // it needs to be consistent for all visitors.
4684                         $this->mRevisionTimestamp = $wgContLang->userAdjust( $timestamp, '' );
4685
4686                         wfProfileOut( __METHOD__ );
4687                 }
4688                 return $this->mRevisionTimestamp;
4689         }
4690
4691         /**
4692          * Mutator for $mDefaultSort
4693          *
4694          * @param $sort New value
4695          */
4696         public function setDefaultSort( $sort ) {
4697                 $this->mDefaultSort = $sort;
4698         }
4699
4700         /**
4701          * Accessor for $mDefaultSort
4702          * Will use the title/prefixed title if none is set
4703          *
4704          * @return string
4705          */
4706         public function getDefaultSort() {
4707                 if( $this->mDefaultSort !== false ) {
4708                         return $this->mDefaultSort;
4709                 } else {
4710                         return $this->mTitle->getNamespace() == NS_CATEGORY
4711                                         ? $this->mTitle->getText()
4712                                         : $this->mTitle->getPrefixedText();
4713                 }
4714         }
4715
4716 }
4717
4718 class OnlyIncludeReplacer {
4719         var $output = '';
4720
4721         function replace( $matches ) {
4722                 if ( substr( $matches[1], -1 ) == "\n" ) {
4723                         $this->output .= substr( $matches[1], 0, -1 );
4724                 } else {
4725                         $this->output .= $matches[1];
4726                 }
4727         }
4728 }
4729
4730 class StripState {
4731         var $general, $nowiki;
4732
4733         function __construct() {
4734                 $this->general = new ReplacementArray;
4735                 $this->nowiki = new ReplacementArray;
4736         }
4737
4738         function unstripGeneral( $text ) {
4739                 wfProfileIn( __METHOD__ );
4740                 $text = $this->general->replace( $text );
4741                 wfProfileOut( __METHOD__ );
4742                 return $text;
4743         }
4744
4745         function unstripNoWiki( $text ) {
4746                 wfProfileIn( __METHOD__ );
4747                 $text = $this->nowiki->replace( $text );
4748                 wfProfileOut( __METHOD__ );
4749                 return $text;
4750         }
4751
4752         function unstripBoth( $text ) {
4753                 wfProfileIn( __METHOD__ );
4754                 $text = $this->general->replace( $text );
4755                 $text = $this->nowiki->replace( $text );
4756                 wfProfileOut( __METHOD__ );
4757                 return $text;
4758         }
4759 }
4760
4761 ?>